- Input $/1M
- $0.30Promo
- Output $/1M
- $1.20Promo
- Context
- 1M
- Max output
- 524K
- Reasoning
- Yes
- Included in
- 8 plans
PromoPermanent 50% discount. The provider's regular rates are listed below.
| Rate | Input | Output | Cache read |
|---|---|---|---|
| Promotional rate | Input$0.30 | Output$1.20 | Cache read$0.06 |
| Promotional rate, above 512K input tokens | Input$0.60 | Output$2.40 | Cache read$0.12 |
| Regular rate | Input$0.60 | Output$2.40 | Cache read$0.12 |
| Regular rate, above 512K input tokens | Input$1.20 | Output$4.80 | Cache read$0.24 |
Published rate checked 2026-09-29. Sources below.
Priority service tier costs 1.5x standard. No cache-write rate is published for M3.
Current published base rates. Context tiers, cache-write duration and other conditions may change the rate for a request. These are not a reconstructed historical invoice.
API token estimate
Token estimate unavailable
This model has conditional rates. Aggregate token totals cannot establish which request or schedule conditions apply.
- Context window
- 1,000,000
- Maximum output
- 524,288
- Input
- text · image · video
- Reasoning
- Supported
- Tool calling
- Supported
- Thinking
- Thinking is off when the thinking parameter is omitted. Set thinking type to adaptive to turn it on.
- Released
- June 1, 2026
- Context
- The API supports up to 1M tokens of context with a guaranteed minimum of 512K tokens.
- Output limit
- max_completion_tokens accepts up to 524,288 tokens; MiniMax recommends 131,072.
- Billing
- Priority service tier costs 1.5x standard.
- Thinking control by model checked Sep 29, 2026
- Context guarantee checked Sep 29, 2026
- Output limit checked Sep 29, 2026
- Discounted rates, long-context tier and priority tier checked Sep 29, 2026
- Release date checked Sep 29, 2026
Benchmarks
Open benchmark sheet →MiniMax · Published Jun 1, 2026
Developer reported · MiniMax M3 release: Terminal-Bench 2.1
Coding
- Terminal-Bench 2.1
- 66.0%
Methodology & sources
MiniMax reports M3 evaluated through its official API using Terminus 2 on internal infrastructure, an 8C16G sandbox, a 2-hour timeout and 128K maximum output tokens.
These scores are reported by MiniMax. Model developers identify authorship; results may originate with another evaluator. Exact benchmark versions remain distinct. StackReplay calculates no composite score.
Terminal-Bench 2.1 · 66.0%
Multi-step tasks in a command-line environment.
Version: 2.1 · Reported benchmark score · percent · Higher is better · Task subset: Not separately specified
MiniMax · Published Jun 1, 2026 · Developer reported
- Reporting source
- MiniMax · MiniMax M3 release: Terminal-Bench 2.1
- Evaluation origin
- MiniMax administered and reported the evaluation (source-computed)
- Effort
- Not reported
- Harness
- Terminus 2
- Tools
- Not reported
- Fallback
- Not reported
- Provider
- MiniMax official API (exact endpoint/version unreported)
- Checked
- Oct 4, 2026
The source explicitly reports 66.0%. Internal infrastructure; sandbox=8C16G; timeout=2 hours; maximum output tokens=128K. The source says models other than its named foreign leaderboard entries were tested via official API; M3 is a MiniMax-tested result. No foreign developer score is admitted. Publication uses the visible article date 2026-06-01. JSON-LD datePublished is 2026-05-31T17:31:18.000Z; both source dates are retained.
Evaluation start, completion and original run publication dates, actual scored task count, task subset, suite revision beyond 2.1, exact model endpoint/version, prompts, seeds and error handling are unreported. Publication dates identify the source articles, not the evaluation runs. The official API evaluation surface is reported; the exact endpoint and model version are unreported. Reasoning effort, tools, trial count, temperature/top-p and the exact Terminus 2 revision are unreported.
$9.99 monthly; weekly and monthly usage windows follow the calendar.
Monthly cycle; purchased credits carry over without expiry.
New subscriptions use per-model allowances from September 28, 2026. Existing flat-$10 subscriptions change at their next renewal.
Monthly cycle; purchased credits carry over without expiry.
Monthly cycle; purchased credits carry over without expiry.
Monthly cycle; purchased credits carry over without expiry.
Monthly credit anniversary; included credits do not carry over.
Monthly subscription; one subscriber per workspace. Paid balance fallback is optional. Recurring charges continue until canceled through account settings. Reset anchors, rollover and a Go-specific refund policy are not established by the reviewed documentation.
Monthly subscription; one subscriber per workspace. Paid balance fallback is optional. Recurring charges continue until canceled through account settings. Reset anchors, rollover and a Go-specific refund policy are not established by the reviewed documentation.
Pricing, assumptions & evidence
Published base rate
Input $0.30 · Output $1.20 · Cache read $0.06 · Cache write Not listed
Above 512K input tokens: input $0.60, output $2.40, cache read $0.12.
- Official context window checked Sep 29, 2026
- Maximum max_completion_tokens checked Sep 29, 2026
- Thinking control and supported message content (text, image, video) by model checked Sep 29, 2026
- Tool use and interleaved thinking checked Sep 29, 2026
- Thinking control by model checked Sep 29, 2026
- Context guarantee checked Sep 29, 2026
- Output limit checked Sep 29, 2026
- Discounted rates, long-context tier and priority tier checked Sep 29, 2026
- Release date checked Sep 29, 2026
- MiniMax-M3 model documentation checked Sep 29, 2026
- MiniMax pay-as-you-go API pricing checked Sep 29, 2026
Missing token-category prices are not zero. Workload pricing applies exact recorded categories and admitted routes. No benchmark score or quality ranking is inferred from prices.
Aliases, routes and identity
Catalog ID minimax-m3
Model release (kind: release)
Developer: MiniMax
minimax/minimax-m3harness_alias