Grok 4.6 pricing: real cost per million tokens
What SpaceXAI's Grok 4.6 really costs across 12 plans from 5 providers: monthly fees, usable allowances and real per-token prices, against the official API list price and independent leaderboards.
Data updated · How real prices are computed
As of 2026-10-08, the cheapest way to use Grok 4.6 is Cursor Ultra at a real cost of about $0.02585 per million tokens ($200/month for roughly 7.737B tokens a month). We track 12 plans and API channels that serve Grok 4.6. Official API list price per million tokens: $0.5 cached input, $2 input, $6 output — about $0.565 per million tokens on a standard workload mix.
- Lowest real price
- $0.02585/MTokCursor Ultra
- API list price
- $0.565/MTokstandard workload mix
- Most monthly tokens
- 7.737BCursor Ultra
- Plans & APIs
- 1211 subscriptions · 1 metered API
Plot Grok 4.6 on the price–capability chart →
Grok 4.6 plans and real prices
Real price = monthly fee ÷ tokens you can actually use in a month. “vs API” shows how many times cheaper than the official API list price (standard workload mix) the plan is. “Evidence” opens the full calculation and sources for that row.
| Plan | Provider | Billing | Monthly fee | Monthly tokens | Real $/MTok | vs API | Confidence | Data date | |
|---|---|---|---|---|---|---|---|---|---|
| Cursor Ultra | Cursor | Subscription | $200/mo | 7.737B | $0.02585 | 21.9× | Medium | Evidence ↓ | |
| Cursor Pro+ | Cursor | Subscription | $60/mo | 2.063B | $0.02908 | 19.4× | Medium | Evidence ↓ | |
| Droid Max | Factory | Subscription | $200/mo | 5.052B | $0.03959 | 14.3× | Medium | Evidence ↓ | |
| Cursor Pro | Cursor | Subscription | $20/mo | 0.47B | $0.04255 | 13.3× | Medium | Evidence ↓ | |
| SuperGrok Plus | SpaceXAI | Subscription | $100/mo | 2.06B | $0.04854 | 11.6× | Medium | Evidence ↓ | |
| SuperGrok | SpaceXAI | Subscription | $30/mo | 0.516B | $0.05814 | 9.7× | High | Evidence ↓ | |
| SuperGrok Heavy | SpaceXAI | Subscription | $300/mo | 5.16B | $0.05814 | 9.7× | Medium | Evidence ↓ | |
| Cursor Ultra (Fast) | Cursor | Subscription | $200/mo | 3.074B | $0.06506 | 8.7× | High | Evidence ↓ | |
| SuperGrok Lite | SpaceXAI | Subscription | $10/mo | 0.15B | $0.06667 | 8.5× | Low | Evidence ↓ | |
| Command Code GOAT | Command Code | Subscription | $10/mo | 0.035B | $0.2825 | 2× | Medium | Evidence ↓ | |
| OpenCode Go | OpenCode | Subscription | $10/mo | 0.027B | $0.3774 | 1.5× | Medium | Evidence ↓ | |
| Grok 4.6 API (<200k) | SpaceXAI | Metered API | — | — | $0.565 | — | High | Evidence ↓ |
Behind each number
For every plan: the data date, the exact calculation, the quota basis, how leaderboard scores were mapped, and the original adoption evidence with its sources.
Cursor UltraCursor$0.0258498/MTok
- Plan
- Cursor Ultra · Cursor
- Data date
- 2026-08-26 – 2026-09-05 · sampled
- Real price / MTok
- $0.0258498
- Monthly tokens
- 7,737,000,000 (7.737B)
- Quota confidence
- Medium
- Calculation
- $200 / month ÷ 7,737,000,000 tokens × 1,000,000 ≈ $0.0258498 / MTok
- Quota basis
Not workload-normalized: sample lacks a token breakdown (or is an official absolute token table); raw tokens used as-is
Official API list (cached / in / out, per MTok): $0.5 / $2 / $6 · standard-load blended ≈ $0.565/MTok
- Benchmark mapping
- Artificial Analysis Intelligence Index v4.3: Medium (4 configurations)
Terminal-Bench 4.0: Low (1 configuration)
Terminal-Bench 4.0 (AA): Low (4 configurations)
Code Arena | WebDev Overall: Low (1 configuration)
Agent Arena · Overall (Ranking / Pareto): Low (1 configuration)
Artificial Analysis Coding Agent Index v1.5: Low (1 configuration)
DeepSWE v1.1: Low (4 configurations)Mapping confidence: how directly each score refers to the model served here.
Adoption evidence · source and rationale (translated from the Chinese record)
Midpoint of two adjusted personal Ultra standard main rows; cursor-adoption-round8-2026-09-06.json
Old 8B→7.737B: (maintainer's current smoothed account 61.0M/0.9% = 6.778B + community 8/26 figure 1478.2M/17% = 8.695B)/2. The community figure may contain heavy first-week half-price usage, so a cost-percentage back-derivation skews high; the midpoint is not a direct single-row measurement — differences in token-type mix and panel rounding are retained; not workload-normalized: the panel reports only token total and cost %
cursor-adoption-round8-2026-09-06.jsonBrowse the public evidence archive
Cursor Pro+Cursor$0.029081/MTok
- Plan
- Cursor Pro+ · Cursor
- Data date
- 2026-08-26 – 2026-09-05 · derived; uses the data date of Cursor Ultra · grok-4.6
- Real price / MTok
- $0.029081
- Monthly tokens
- 2,063,200,000 (2.063B)
- Quota confidence
- Medium
- Calculation
- $60 / month ÷ 2,063,200,000 tokens × 1,000,000 ≈ $0.029081 / MTok
- Quota basis
Not workload-normalized: sample lacks a token breakdown (or is an official absolute token table); raw tokens used as-is
Official API list (cached / in / out, per MTok): $0.5 / $2 / $6 · standard-load blended ≈ $0.565/MTok
- Benchmark mapping
- Artificial Analysis Intelligence Index v4.3: Medium (4 configurations)
Terminal-Bench 4.0: Low (1 configuration)
Terminal-Bench 4.0 (AA): Low (4 configurations)
Code Arena | WebDev Overall: Low (1 configuration)
Agent Arena · Overall (Ranking / Pareto): Low (1 configuration)
Artificial Analysis Coding Agent Index v1.5: Low (1 configuration)
DeepSWE v1.1: Low (4 configurations)Mapping confidence: how directly each score refers to the model served here.
Adoption evidence · source and rationale (translated from the Chinese record)
round3 panel Pro+ pool ≈$800; scaled in proportion to the Ultra pool $3000; cursor-adoption-round8-2026-09-06.json
Old 2.133B→2.063B: 77.37×800/3000; inherits the cross-tier pool-size assumption, not an independent measurement; the community-figure back-derivation of $4500~4800 was not adopted as the official pool; promotion and account differences retained
cursor-adoption-round8-2026-09-06.jsonBrowse the public evidence archive
Droid MaxFactory$0.0395883/MTok
- Data date
- 2026-09-29 · derived; uses the data date of Droid Max · claude-opus-5.5
- Real price / MTok
- $0.0395883
- Monthly tokens
- 5,052,000,000 (5.052B)
- Quota confidence
- Medium
- Calculation
- $200 / month ÷ 5,052,000,000 tokens × 1,000,000 ≈ $0.0395883 / MTok
- Quota basis
Anthropic workload: 97% cache reads / 2.5% cache writes / 0.5% output
Official API list (cached / in / out, per MTok): $0.5 / $2 / $6 · standard-load blended ≈ $0.565/MTok
- Benchmark mapping
- Artificial Analysis Intelligence Index v4.3: Medium (4 configurations)
Terminal-Bench 4.0: Low (1 configuration)
Terminal-Bench 4.0 (AA): Low (4 configurations)
Code Arena | WebDev Overall: Low (1 configuration)
Agent Arena · Overall (Ranking / Pareto): Low (1 configuration)
Artificial Analysis Coding Agent Index v1.5: Low (1 configuration)
DeepSWE v1.1: Low (4 configurations)Mapping confidence: how directly each score refers to the model served here.
Adoption evidence · source and rationale (translated from the Chinese record)
Derived from same-plan claude-opus-5.5 2.526B × 2.0
Factory official multipliers Opus 5.5 1.6× / grok-4.6 0.8×: the same Standard Usage pool scaled by multiplier — not a measurement of this model; inherits the Opus 5.5 row's Anthropic-tier workload, not recomputed for each model's own price gap; docs.factory.ai/docs/models; droid-model-multipliers-2026-09-29.json
droid-model-multipliers-2026-09-29.jsonBrowse the public evidence archive
Cursor ProCursor$0.0425532/MTok
- Plan
- Cursor Pro · Cursor
- Data date
- 2026-08-27 · sampled
- Real price / MTok
- $0.0425532
- Monthly tokens
- 470,000,000 (0.47B)
- Quota confidence
- Medium
- Calculation
- $20 / month ÷ 470,000,000 tokens × 1,000,000 ≈ $0.0425532 / MTok
- Quota basis
Not workload-normalized: sample lacks a token breakdown (or is an official absolute token table); raw tokens used as-is
Official API list (cached / in / out, per MTok): $0.5 / $2 / $6 · standard-load blended ≈ $0.565/MTok
- Benchmark mapping
- Artificial Analysis Intelligence Index v4.3: Medium (4 configurations)
Terminal-Bench 4.0: Low (1 configuration)
Terminal-Bench 4.0 (AA): Low (4 configurations)
Code Arena | WebDev Overall: Low (1 configuration)
Agent Arena · Overall (Ranking / Pareto): Low (1 configuration)
Artificial Analysis Coding Agent Index v1.5: Low (1 configuration)
DeepSWE v1.1: Low (4 configurations)Mapping confidence: how directly each score refers to the model served here.
Adoption evidence · source and rationale (translated from the Chinese record)
Cursor forum panel: 303.9M = 65% → 468M; a separate user's verbal report of 400M~500M to saturation
Kept the independent panel adoption of 470M, not linked to the Ultra midpoint; the pool is denominated in compute cost, not raw tokens; not workload-normalized: the panel reports only token total and cost %
SuperGrok PlusSpaceXAI$0.0485437/MTok
- Plan
- SuperGrok Plus · SpaceXAI
- Data date
- 2026-08-23 – 2026-09-20 · derived; uses the data date of SuperGrok · grok-4.6
- Real price / MTok
- $0.0485437
- Monthly tokens
- 2,060,000,000 (2.06B)
- Quota confidence
- Medium
- Calculation
- $100 / month ÷ 2,060,000,000 tokens × 1,000,000 ≈ $0.0485437 / MTok
- Quota basis
Standard workload: 97% cache reads / 2.5% input / 0.5% output
Official API list (cached / in / out, per MTok): $0.5 / $2 / $6 · standard-load blended ≈ $0.565/MTok
- Benchmark mapping
- Artificial Analysis Intelligence Index v4.3: Medium (4 configurations)
Terminal-Bench 4.0: Low (1 configuration)
Terminal-Bench 4.0 (AA): Low (4 configurations)
Code Arena | WebDev Overall: Low (1 configuration)
Agent Arena · Overall (Ranking / Pareto): Low (1 configuration)
Artificial Analysis Coding Agent Index v1.5: Low (1 configuration)
DeepSWE v1.1: Low (4 configurations)Mapping confidence: how directly each score refers to the model served here.
Adoption evidence · source and rationale (translated from the Chinese record)
Panel weekly quota $100 × Super's precise calibration ×4 weeks
Adopted 2.06B: 128,937,168×4 weeks×100/25, taken to one decimal; a linux.do user's report that every dollar of usage raises the bar 1% agrees with $100/week
SuperGrokSpaceXAI$0.0581395/MTok
- Data date
- 2026-08-23 – 2026-09-20 · sampled
- Real price / MTok
- $0.0581395
- Monthly tokens
- 516,000,000 (0.516B)
- Quota confidence
- High
- Calculation
- $30 / month ÷ 516,000,000 tokens × 1,000,000 ≈ $0.0581395 / MTok
- Quota basis
Standard workload: 97% cache reads / 2.5% input / 0.5% output
Official API list (cached / in / out, per MTok): $0.5 / $2 / $6 · standard-load blended ≈ $0.565/MTok
- Benchmark mapping
- Artificial Analysis Intelligence Index v4.3: Medium (4 configurations)
Terminal-Bench 4.0: Low (1 configuration)
Terminal-Bench 4.0 (AA): Low (4 configurations)
Code Arena | WebDev Overall: Low (1 configuration)
Agent Arena · Overall (Ranking / Pareto): Low (1 configuration)
Artificial Analysis Coding Agent Index v1.5: Low (1 configuration)
DeepSWE v1.1: Low (4 configurations)Mapping confidence: how directly each score refers to the model served here.
Adoption evidence · source and rationale (translated from the Chinese record)
V2EX controlled saturation 127,272,629 tokens = 100% of a full week, merged per percentage point with issue #56 (Grok Build _x.ai/billing polling) two ordinary weeks 223.1M = +193pp → weekly pool 128,937,168; community-issues-round1-2026-09-29.json; workload-conversion-round1-2026-09-30.json
478M→516M (2026-09-30 decision: uniform workload-normalization for measured samples with token breakdown): the V2EX segment has no breakdown and is treated as already standard-tier for worth (127,272,629×$0.565 = $71.909) + #56 segment worth $141.54 (cache 206.4M×$0.5 + in 15.3M×$2 + out 1.29M×$6) ÷293pp = $0.7284/pp → $72.85/week ×4 weeks ÷ standard-tier blended price $0.565/MTok = weekly pool 128,937,168; the raw-basis 4.78 is kept for comparison; suspected double-activity weeks (236M, 274M) and low-hit weeks were excluded. Original 2026-09-29 adoption decision (raw basis) — 509M→478M, 2026-09-29 decision: issue #56 (Grok Build _x.ai/billing weekly-quota % every 5 minutes, cross-checked against unified.jsonl) two full ordinary weeks 223.1M raw = +193 percentage points (+94%/120.9M, +99%/102.2M; 4.62 alone), merged per percentage point with the V2EX controlled saturation 127,272,629 tokens (100% of a week): (127.272629+223.1)÷(100+193) = 1.1958M/pp ×100×4 weeks = 478M. Evidence: community-issues-round1-2026-09-29.json.
community-issues-round1-2026-09-29.jsonworkload-conversion-round1-2026-09-30.jsonBrowse the public evidence archive
SuperGrok HeavySpaceXAI$0.0581395/MTok
- Plan
- SuperGrok Heavy · SpaceXAI
- Data date
- 2026-08-23 – 2026-09-20 · derived; uses the data date of SuperGrok · grok-4.6
- Real price / MTok
- $0.0581395
- Monthly tokens
- 5,160,000,000 (5.16B)
- Quota confidence
- Medium
- Calculation
- $300 / month ÷ 5,160,000,000 tokens × 1,000,000 ≈ $0.0581395 / MTok
- Quota basis
Standard workload: 97% cache reads / 2.5% input / 0.5% output
Official API list (cached / in / out, per MTok): $0.5 / $2 / $6 · standard-load blended ≈ $0.565/MTok
- Benchmark mapping
- Artificial Analysis Intelligence Index v4.3: Medium (4 configurations)
Terminal-Bench 4.0: Low (1 configuration)
Terminal-Bench 4.0 (AA): Low (4 configurations)
Code Arena | WebDev Overall: Low (1 configuration)
Agent Arena · Overall (Ranking / Pareto): Low (1 configuration)
Artificial Analysis Coding Agent Index v1.5: Low (1 configuration)
DeepSWE v1.1: Low (4 configurations)Mapping confidence: how directly each score refers to the model served here.
Adoption evidence · source and rationale (translated from the Chinese record)
Panel weekly quota $250 × Super's precise calibration ×4 weeks
Adopted 5.16B: 128,937,168×4 weeks×250/25, taken to one decimal; the list-price conversion 1.8B voided (panel dollars ≠ list-price dollars); Zhang's 20.8B not adopted
Cursor Ultra (Fast)Cursor$0.0650618/MTok
- Plan
- Cursor Ultra (Fast) · Cursor
- Data date
- 2026-09-05 · sampled
- Real price / MTok
- $0.0650618
- Monthly tokens
- 3,074,000,000 (3.074B)
- Quota confidence
- High
- Calculation
- $200 / month ÷ 3,074,000,000 tokens × 1,000,000 ≈ $0.0650618 / MTok
- Quota basis
Not workload-normalized: sample lacks a token breakdown (or is an official absolute token table); raw tokens used as-is
Official API list (cached / in / out, per MTok): $0.5 / $2 / $6 · standard-load blended ≈ $0.565/MTok
- Benchmark mapping
- Artificial Analysis Intelligence Index v4.3: Medium (4 configurations)
Terminal-Bench 4.0: Low (1 configuration)
Terminal-Bench 4.0 (AA): Low (4 configurations)
Code Arena | WebDev Overall: Low (1 configuration)
Agent Arena · Overall (Ranking / Pareto): Low (1 configuration)
Artificial Analysis Coding Agent Index v1.5: Low (1 configuration)
DeepSWE v1.1: Low (4 configurations)Mapping confidence: how directly each score refers to the model served here.
Adoption evidence · source and rationale (translated from the Chinese record)
Direct back-derivation from the maintainer's current smoothed-account screenshot 863.8M/28.1%; cursor-adoption-round8-2026-09-06.json
Old 4B→3.074B; direct back-derivation from the largest sample's xhigh-fast row, percentage-rounding interval 3.069B~3.080B; the same figure's smaller high-fast row 2.443B not adopted. Standard/Fast are not required to be a strict 2× in raw tokens, because the panel deducts by cost and the token-type mix differs; the official three-part 2× rate fact is unchanged; kept separate from the SuperGrok channel; not workload-normalized: the panel reports only token total and cost %
cursor-adoption-round8-2026-09-06.jsonBrowse the public evidence archive
SuperGrok LiteSpaceXAI$0.0666667/MTok
- Plan
- SuperGrok Lite · SpaceXAI
- Data date
- 2026-07 · sampled
- Real price / MTok
- $0.0666667
- Monthly tokens
- 150,000,000 (0.15B)
- Quota confidence
- Low
- Calculation
- $10 / month ÷ 150,000,000 tokens × 1,000,000 ≈ $0.0666667 / MTok
- Quota basis
Not workload-normalized: sample lacks a token breakdown (or is an official absolute token table); raw tokens used as-is
Official API list (cached / in / out, per MTok): $0.5 / $2 / $6 · standard-load blended ≈ $0.565/MTok
- Benchmark mapping
- Artificial Analysis Intelligence Index v4.3: Medium (4 configurations)
Terminal-Bench 4.0: Low (1 configuration)
Terminal-Bench 4.0 (AA): Low (4 configurations)
Code Arena | WebDev Overall: Low (1 configuration)
Agent Arena · Overall (Ranking / Pareto): Low (1 configuration)
Artificial Analysis Coding Agent Index v1.5: Low (1 configuration)
DeepSWE v1.1: Low (4 configurations)Mapping confidence: how directly each score refers to the model served here.
Adoption evidence · source and rationale (translated from the Chinese record)
aa_grok_build_2026_07
Panel weekly quota unknown — three rounds of web search found nothing; not workload-normalized: AA placeholder value, no measured sample
Command Code GOATCommand Code$0.282486/MTok
- Data date
- 2026-09-30 · official source date
- Real price / MTok
- $0.282486
- Monthly tokens
- 35,400,000 (0.035B)
- Quota confidence
- Medium
- Calculation
- $10 / month ÷ 35,400,000 tokens × 1,000,000 ≈ $0.282486 / MTok
- Quota basis
Standard workload: 97% cache reads / 2.5% input / 0.5% output
Official API list (cached / in / out, per MTok): $0.5 / $2 / $6 · standard-load blended ≈ $0.565/MTok
- Benchmark mapping
- Artificial Analysis Intelligence Index v4.3: Medium (4 configurations)
Terminal-Bench 4.0: Low (1 configuration)
Terminal-Bench 4.0 (AA): Low (4 configurations)
Code Arena | WebDev Overall: Low (1 configuration)
Agent Arena · Overall (Ranking / Pareto): Low (1 configuration)
Artificial Analysis Coding Agent Index v1.5: Low (1 configuration)
DeepSWE v1.1: Low (4 configurations)Mapping confidence: how directly each score refers to the model served here.
Adoption evidence · source and rationale (translated from the Chinese record)
https://commandcode.ai/docs/plans/goat official per-model allowance and three-part pricing; https://commandcode.ai/pricing; $10->$70 credits; code-subscriptions-round1-2026-09-06.json; goat-opencode-catalogs-round3-2026-09-30.json
Added 35.4M: min(shared monthly pool $70, model allowance $20) ÷ project standard-workload weighted price; official three-part pricing; New models default $20. Processing fee ignored; allowances of different models in the same plan are not additive; no usage-panel token+% screenshot — computed from official absolute credits and the price list
https://commandcode.ai/docs/plans/goathttps://commandcode.ai/pricing;code-subscriptions-round1-2026-09-06.jsongoat-opencode-catalogs-round3-2026-09-30.jsonBrowse the public evidence archive
OpenCode GoOpenCode$0.377358/MTok
- Plan
- OpenCode Go · OpenCode
- Data date
- 2026-09-30 · official source date
- Real price / MTok
- $0.377358
- Monthly tokens
- 26,500,000 (0.027B)
- Quota confidence
- Medium
- Calculation
- $10 / month ÷ 26,500,000 tokens × 1,000,000 ≈ $0.377358 / MTok
- Quota basis
Standard workload: 97% cache reads / 2.5% input / 0.5% output
Official API list (cached / in / out, per MTok): $0.5 / $2 / $6 · standard-load blended ≈ $0.565/MTok
- Benchmark mapping
- Artificial Analysis Intelligence Index v4.3: Medium (4 configurations)
Terminal-Bench 4.0: Low (1 configuration)
Terminal-Bench 4.0 (AA): Low (4 configurations)
Code Arena | WebDev Overall: Low (1 configuration)
Agent Arena · Overall (Ranking / Pareto): Low (1 configuration)
Artificial Analysis Coding Agent Index v1.5: Low (1 configuration)
DeepSWE v1.1: Low (4 configurations)Mapping confidence: how directly each score refers to the model served here.
Adoption evidence · source and rationale (translated from the Chinese record)
https://opencode.ai/docs/go/ official per-model Usage and three-part pricing; opencode-go-round5-2026-09-06.json; goat-opencode-catalogs-round3-2026-09-30.json
Old: 27.9M (request estimate)→26.5M: min(shared monthly pool $60, model Usage $15) ÷ project standard-workload weighted price; ≤200K list price; the >200K price doubles, kept in research variants. The official request counts serve only as a cross-check and are no longer the primary allowance figure; allowances of different models in the same plan are not additive
https://opencode.ai/docs/go/goat-opencode-catalogs-round3-2026-09-30.jsonopencode-go-round5-2026-09-06.jsonBrowse the public evidence archive
Grok 4.6 API (<200k)SpaceXAI$0.565/MTok
- Plan
- Grok 4.6 API (<200k) · SpaceXAI
- Data date
- 2026-09-05 · official source date
- Real price / MTok
- $0.565
- Quota confidence
- High
- Calculation
- Metered API · public token prices weighted by the project standard workload.
- Benchmark mapping
- Artificial Analysis Intelligence Index v4.3: Medium (4 configurations)
Terminal-Bench 4.0: Low (1 configuration)
Terminal-Bench 4.0 (AA): Low (4 configurations)
Code Arena | WebDev Overall: Low (1 configuration)
Agent Arena · Overall (Ranking / Pareto): Low (1 configuration)
Artificial Analysis Coding Agent Index v1.5: Low (1 configuration)
DeepSWE v1.1: Low (4 configurations)Mapping confidence: how directly each score refers to the model served here.
Adoption evidence · source and rationale (translated from the Chinese record)
docs.x.ai
List price cached 0.5/in 2.0/out 6.0 × project standard workload 97.0%/2.50%/0.50%
Grok 4.6 benchmark scores
Highest archived configuration per leaderboard; ranks are among all models each board archives in this dataset. Scores are never mixed across boards.
| Leaderboard | Metric | Score | Rank | Configuration | Effort | Harness | Configs | Snapshot |
|---|---|---|---|---|---|---|---|---|
| Artificial Analysis Intelligence Index v4.3 | Intelligence Index | 44.31 | #15 / 60 | Grok 4.6 (high) | High | — | 4 | 2026-10-01 |
| Terminal-Bench 4.0 | Resolution Rate % | 20.3 | #13 / 18 | Grok Build - Grok 4.6 (high) | High | Grok Build | 1 | 2026-09-21 |
| Terminal-Bench 4.0 (AA) | Resolution Rate % | 21.21 | #21 / 53 | Artificial Analysis - Grok 4.6 (high) | High | Artificial Analysis | 4 | 2026-10-01 |
| Code Arena | WebDev Overall | Arena Score | 1,620 | #16 / 54 | grok-4.6-high | High | — | 1 | 2026-09-30 |
| Agent Arena · Overall (Ranking / Pareto) | Net Improvement % | 1.16 | #22 / 39 | Grok 4.6 (xHigh) | xhigh | — | 1 | 2026-09-30 |
| Artificial Analysis Coding Agent Index v1.5 | Coding Agent Index | 46.97 | #13 / 19 | Grok Build - Grok 4.6 (xhigh) | xhigh | Grok Build | 1 | 2026-10-01 |
| DeepSWE v1.1 | Pass@1 % | 67.48 | #13 / 33 | mini-swe-agent - grok-4-6 (medium) | Medium | mini-swe-agent | 4 | 2026-09-03 |
Every archived configuration
Artificial Analysis Intelligence Index v4.3 4 configurations
| Configuration | Score | Score interval | Harness / effort / mode | Source task cost (mean / median) | Checked | Links |
|---|---|---|---|---|---|---|
| Grok 4.6 (high) | 44.3113 | — | — / High / — | — / — | 2026-10-01 | Source · Archive |
| Grok 4.6 (xhigh) | 44.1998 | — | — / xhigh / — | — / — | 2026-10-01 | Source · Archive |
| Grok 4.6 (medium) | 42.8363 | — | — / Medium / — | — / — | 2026-10-01 | Source · Archive |
| Grok 4.6 (low) | 35.1247 | — | — / Low / — | — / — | 2026-10-01 | Source · Archive |
Terminal-Bench 4.0 1 configuration
| Configuration | Score | Score interval | Harness / effort / mode | Source task cost (mean / median) | Checked | Links |
|---|---|---|---|---|---|---|
| Grok Build - Grok 4.6 (high)Official Terminal-Bench 4.0 leaderboard row; score belongs to this harness × model × effort configuration. | 20.3 | 17.21 – 23.39 | Grok Build / High / — | — / — | 2026-10-01 | Source · Archive |
Terminal-Bench 4.0 (AA) 4 configurations
| Configuration | Score | Score interval | Harness / effort / mode | Source task cost (mean / median) | Checked | Links |
|---|---|---|---|---|---|---|
| Artificial Analysis - Grok 4.6 (high)AA's own-harness independent measurement of Terminal-Bench 4.0 (the terminalBench40 field in the models-board payload, fraction×100); scored separately from the official tbench.ai board. | 21.2121 | — | Artificial Analysis / High / — | — / — | 2026-10-01 | Source · Archive |
| Artificial Analysis - Grok 4.6 (xhigh)AA's own-harness independent measurement of Terminal-Bench 4.0 (the terminalBench40 field in the models-board payload, fraction×100); scored separately from the official tbench.ai board. | 17.1717 | — | Artificial Analysis / xhigh / — | — / — | 2026-10-01 | Source · Archive |
| Artificial Analysis - Grok 4.6 (medium)AA's own-harness independent measurement of Terminal-Bench 4.0 (the terminalBench40 field in the models-board payload, fraction×100); scored separately from the official tbench.ai board. | 13.1313 | — | Artificial Analysis / Medium / — | — / — | 2026-10-01 | Source · Archive |
| Artificial Analysis - Grok 4.6 (low)AA's own-harness independent measurement of Terminal-Bench 4.0 (the terminalBench40 field in the models-board payload, fraction×100); scored separately from the official tbench.ai board. | 3.0303 | — | Artificial Analysis / Low / — | — / — | 2026-10-01 | Source · Archive |
Code Arena | WebDev Overall 1 configuration
Agent Arena · Overall (Ranking / Pareto) 1 configuration
Artificial Analysis Coding Agent Index v1.5 1 configuration
| Configuration | Score | Score interval | Harness / effort / mode | Source task cost (mean / median) | Checked | Links |
|---|---|---|---|---|---|---|
| Grok Build - Grok 4.6 (xhigh)Score belongs to this named agent/harness × model configuration; it is not a harness-independent model measurement. indexScore stored as percent (raw fraction × 100). | 46.969 | — | Grok Build / xhigh / — | $3.57 / — | 2026-10-01 | Source · Archive |
DeepSWE v1.1 4 configurations
| Configuration | Score | Score interval | Harness / effort / mode | Source task cost (mean / median) | Checked | Links |
|---|---|---|---|---|---|---|
| mini-swe-agent - grok-4-6 (medium) | 67.4779 | 65.197 – 69.759 | mini-swe-agent / Medium / — | $3.449 / $2.946 | 2026-09-12 | Source · Archive |
| mini-swe-agent - grok-4-6 (xhigh) | 66.7406 | 64.565 – 68.916 | mini-swe-agent / xhigh / — | $5.498 / $4.639 | 2026-09-12 | Source · Archive |
| mini-swe-agent - grok-4-6 (high) | 65.1885 | 63.655 – 66.722 | mini-swe-agent / High / — | $4.385 / $3.694 | 2026-09-12 | Source · Archive |
| mini-swe-agent - grok-4-6 (low) | 41.6481 | 39.329 – 43.967 | mini-swe-agent / Low / — | $1.042 / $0.7426 | 2026-09-12 | Source · Archive |
These are reference mappings, not benchmarks of a subscription or API channel. Harness and quota-measurement effort alignment are unverified. Source task costs are not subscription task costs.
- Exact served-model reference; quota-measurement effort is unverified.
- Exact served-model reference only; product harness and quota-measurement effort are unverified. Not a benchmark measurement of this subscription or API channel.
Grok 4.6 official API pricing
| Cached input | Input | Output |
|---|---|---|
| $0.5 | $2 | $6 |
Frequently asked questions
How much does Grok 4.6 really cost per million tokens?
Measured as monthly fee ÷ tokens you can actually use, Grok 4.6 ranges from $0.02585 (Cursor Ultra) to $0.565 (Grok 4.6 API (<200k)) per million tokens.
Which subscription gives the most Grok 4.6 usage?
Cursor Ultra allows roughly 7.737B tokens of Grok 4.6 per month for $200.
What is the official API price of Grok 4.6?
Official API list price per million tokens: $0.5 cached input, $2 input, $6 output — about $0.565 per million tokens on a standard workload mix.
How does Grok 4.6 score on independent leaderboards?
Grok 4.6 has scores on 7 leaderboards; its strongest placement is 44.31 (Intelligence Index) on Artificial Analysis Intelligence Index v4.3, #15 of 60 models tracked.
Who makes Grok 4.6, and where can I use it?
Grok 4.6 (model ID grok-4.6) is made by SpaceXAI and is available through Command Code, Cursor, Factory, OpenCode, SpaceXAI.
More SpaceXAI models
| Model | Lowest real $/MTok | Cheapest plan | Plans |
|---|---|---|---|
| Grok 4.5 | $0.02585 | Cursor Ultra | 4 |
| Grok 4.7 | $0.03959 | Droid Max | 7 |