Mistral AI model product
Pixtral 12B is a multimodal model with a 12B-parameter language model and a 400M-parameter vision encoder.
Updated Sep 8, 2026. Default version: Pixtral-12B
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
Benchmark scores for Pixtral-12B.
Benchmark | Score | Rank | Participants | Percentile | Evidence | Evaluated |
|---|
| BenchmarkMM IF-Eval | Score52.70% | Rank02 | Participants2 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkVQAv2 | Score78.60% | Rank02 | Participants3 | Percentile50.00% | EvidenceC | Evaluated |
| BenchmarkMT-Bench | Score76.80% | Rank11 | Participants12 | Percentile9.09% | EvidenceC | Evaluated |
| BenchmarkMM-MT-Bench | Score0.61 points | Rank13 | Participants17 | Percentile25.00% | EvidenceC | Evaluated |
| BenchmarkChartQA | Score81.80% | Rank18 | Participants26 | Percentile32.00% | EvidenceC | Evaluated |
| BenchmarkDocVQA | Score90.70% | Rank20 | Participants28 | Percentile29.63% | EvidenceC | Evaluated |
| BenchmarkMathVista | Score58.00% | Rank28 | Participants39 | Percentile28.95% | EvidenceC | Evaluated |
| BenchmarkHumanEval | Score72.00% | Rank55 | Participants66 | Percentile16.92% | EvidenceC | Evaluated |
| BenchmarkMMMU | Score52.50% | Rank57 | Participants64 | Percentile11.11% | EvidenceC | Evaluated |
| BenchmarkMATH | Score48.10% | Rank58 | Participants71 | Percentile18.57% | EvidenceC | Evaluated |
| BenchmarkIFEval | Score61.30% | Rank67 | Participants68 | Percentile1.49% | EvidenceC | Evaluated |
| BenchmarkMMLU | Score69.20% | Rank86 | Participants101 | Percentile15.00% | EvidenceC | Evaluated |
| BenchmarkLM Arena Vision | Score1,008.57 rating | Rank99 | Participants107 | Percentile7.55% | EvidenceA | Evaluated |
| BenchmarkLM Arena Vision Style Control | Score1,025.42 rating | Rank100 | Participants107 | Percentile6.60% | EvidenceA | Evaluated |
Preference and agent-evaluation results for the default version.
Arena | Category | Rank | Rating / score | Votes | Observations | Result date |
|---|
| Arenavision | Categorychinese | Rank90 | Rating / score949.66 | Votes335 | ObservationsN/A | Result date |
| Arenavision style control | Categorychinese | Rank92 | Rating / score966.07 | Votes335 | ObservationsN/A | Result date |
| Arenavision | Categoryenglish | Rank98 | Rating / score1,026.40 | Votes4,090 | ObservationsN/A | Result date |
| Arenavision | Categoryoverall | Rank99 | Rating / score1,008.57 | Votes7,511 | ObservationsN/A | Result date |
| Arenavision style control | Categoryenglish | Rank99 | Rating / score1,041.17 | Votes4,090 | ObservationsN/A | Result date |
| Arenavision style control | Categoryoverall | Rank100 | Rating / score1,025.42 | Votes7,511 | ObservationsN/A | Result date |
Official vendor API pricing appears first, followed by individual provider offers.
Provider | Provider model ID | Region | Input / 1M | Output / 1M | Context | Updated |
|---|
| ProviderMistral | Provider model IDpixtral-12b | Regionglobal | Input / 1M$0.15 | Output / 1M$0.15 | Context128K | Updated |
| ProviderScaleway | Provider model IDpixtral-12b-2409 | Regionglobal | Input / 1M$0.20 | Output / 1M$0.20 | Context128K | Updated |
| ProviderCortecs | Provider model IDpixtral-12b-2409 | Regionglobal | Input / 1M$0.223 | Output / 1M$0.223 | Context128K | Updated |
| ProviderGreenPT | Provider model IDpixtral-12b-2409 | Regionglobal | Input / 1M$0.285 | Output / 1M$0.285 | Context128K | Updated |
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
Provider-specific output speed and catalog latency for Pixtral-12B. Runtime does not affect the capability score.
Provider | Output Speed | Catalog Latency | Max Input | Max Output | Updated |
|---|
| ProviderMistral AI | Output Speed0.10 tok/s | Catalog Latency0.50 s | Max Input128K | Max Output8.2K | Updated |
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
Technical details for the model's default version.
Available versions of this model. The score column identifies the version used in the overall ranking.
Version | Released | LLMBoard | Parameters | Context | Max output | Open weights | License |
|---|
| VersionPixtral-12B | Released | LLMBoard0.00 | Parameters12.4B | Context128K | Max output8.2K | Open weightsYes | LicenseApache 2.0 |
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
Key information about Pixtral 12B and its available data.
Pixtral 12B is a multimodal large language model from Mistral AI that understands natural images and documents. It supports variable image sizes and multiple images in context, while also providing text-only performance.
Data as of 2026-09-08.
Common questions about Pixtral 12B.
Pixtral 12B's default version was released on Sep 17, 2024.
Pixtral 12B's official API price is $0.15 per million input tokens and $0.15 per million output tokens via Mistral. The lowest tracked third-party offer starts at $0.20 input and $0.20 output via Scaleway.
Pixtral 12B was created by Mistral AI.
The default version has a 128K token context window.
Yes. The default version is marked as open weight under Apache 2.0.
4 provider offerings are linked to the default version.