Devstral 2 123B Instruct 2512 vs pi 0.7
At a Glance
| Compare |  Devstral 2 123B Instruct 2512Mistral AI |  pi 0.7Physical Intelligence |
|---|---|---|
| Pricing and Limits | ||
| Context windowMaximum documented tokens | 262K | Not reported |
| Model facts checked | Aug 28, 2026View model evidence → | Aug 29, 2026View model evidence → |
Available Benchmarks
Side-by-Side Facts
| Field | Devstral-2-123B-Instruct-2512 | pi 0.7 |
|---|---|---|
| Developer | Mistral AI | Physical Intelligence |
| Family | Devstral 2 123b Instruct 2512 | pi |
| Model | Devstral-2-123B-Instruct-2512 | pi 0.7 |
| Version | Devstral-2-123B-Instruct-2512 | 0.7 |
| Lifecycle | active | active |
| Released | Unknown | 2026-04-16 |
| Knowledge cutoff | Unknown | Unknown |
| Input modalities | Text | Text, Image, Robot state |
| Output modalities | Text | Robot action |
| Context window | 262K | Unknown |
| Total parameters | 125B | Unknown |
| Active parameters | Unknown | Unknown |
| License | other | Unknown |
| Open weights | Yes | No |
| API available | Yes | Unknown |
| Self-hostable | Yes | Unknown |
| Provider access | Openrouter (Standard) | Unknown |
| Capabilities | chat, generation, tools | cross-embodiment, dexterous-manipulation, language-steering, visual-subgoals |
| Robotics model type | Unknown | Vision-language-action model |
| Action representation | Unknown | Continuous robot actions conditioned by multimodal prompts |
| Control architecture | Unknown | High-level policy, world model, and action expert |
| Inference location | Unknown | Unknown |
| Native control rate (Hz) | Unknown | Unknown |
| Supported embodiments | Unknown | mobile manipulators, bimanual UR5e, multiple fixed manipulators |
| Training data | Unknown | Robot demonstrations, autonomous data, egocentric human data, and multimodal web data described by the publisher. |
Devstral 2 123B Instruct 2512 Capabilities
pi 0.7 Capabilities
Related Comparisons
| A | Pair | B | Context |
|---|---|---|---|
Devstral-2-123B-Instruct-2512Mistral AI | vs |  GPT-6 AstraOpenAI | cross-developer peerstext |
Claude Fable 5.1Anthropic | vs | Devstral-2-123B-Instruct-2512Mistral AI | cross-developer peerstext |
 Gemini 3.1 ProGoogle DeepMind | vs | Devstral-2-123B-Instruct-2512Mistral AI | cross-developer peerstext |
 DeepSeek-V4.1-FlashDeepSeek | vs | Devstral-2-123B-Instruct-2512Mistral AI | cross-developer peerstext |
Devstral-2-123B-Instruct-2512Mistral AI | vs | Grok 4.6xAI | cross-developer peerstext |
Devstral-2-123B-Instruct-2512Mistral AI | vs |  Qwen3.8-MaxQwen | cross-developer peerstext |
Devstral-2-123B-Instruct-2512Mistral AI | vs |  Kimi-K3Moonshot AI | cross-developer peerstext |
 MiniMax-M3MiniMax | vs | Devstral-2-123B-Instruct-2512Mistral AI | cross-developer peerstext |
Devstral-2-123B-Instruct-2512Mistral AI | vs |  GLM-5.3Z.ai | cross-developer peerstext |
Devstral-2-123B-Instruct-2512Mistral AI | vs |  Hy4 previewTencent | cross-developer peerstext |
Seed 2.1 ProByteDance Seed | vs | Devstral-2-123B-Instruct-2512Mistral AI | cross-developer peerstext |
Devstral-2-123B-Instruct-2512Mistral AI | vs | Mistral Large 3Mistral AI | developer peerstext |
Primary Evidence
Sources and Freshness
Questions
Devstral 2 123B Instruct 2512 vs pi 0.7 FAQs
Is Devstral 2 123B Instruct 2512 or pi 0.7 better for coding?+
This comparison does not currently contain a protocol-matched coding benchmark for both Devstral 2 123B Instruct 2512 and pi 0.7, so Model Markets cannot name a coding leader from pricing, context size, or capability labels alone.
Which is cheaper, Devstral 2 123B Instruct 2512 or pi 0.7?+
Only Devstral 2 123B Instruct 2512 has a directly sourced input price: $0.44 per million tokens. Only Devstral 2 123B Instruct 2512 has a directly sourced output price: $2.20 per million tokens.
Which has a larger context window, Devstral 2 123B Instruct 2512 or pi 0.7?+
Neither model has a larger sourced context window in this comparison. Devstral 2 123B Instruct 2512 is 262K and pi 0.7 is —.
Which performs better in benchmarks, Devstral 2 123B Instruct 2512 or pi 0.7?+
There is no overall benchmark winner: At least two independently verified, protocol-matched benchmarks are required for an overall winner.
Can Devstral 2 123B Instruct 2512 or pi 0.7 be self-hosted?+
Devstral 2 123B Instruct 2512 is the only model in this pair currently marked as self-hostable. Devstral 2 123B Instruct 2512 is open weight; pi 0.7 is not marked open weight.
Can Devstral 2 123B Instruct 2512 and pi 0.7 understand images?+
Devstral 2 123B Instruct 2512 is not documented with image input; pi 0.7 is documented with image input. This reflects supported input modalities, not vision quality.
Which can generate longer answers, Devstral 2 123B Instruct 2512 or pi 0.7?+
Neither has a larger sourced maximum output. Devstral 2 123B Instruct 2512 is — and pi 0.7 is —.
Do Devstral 2 123B Instruct 2512 and pi 0.7 support reasoning and tool use?+
Devstral 2 123B Instruct 2512: tool calling. pi 0.7: image input. Feature support does not establish relative quality.
Which is available from more inference providers, Devstral 2 123B Instruct 2512 or pi 0.7?+
Devstral 2 123B Instruct 2512 has 1 sourced provider route; pi 0.7 has 0, so Devstral 2 123B Instruct 2512 has broader tracked availability.
Which offers better value, Devstral 2 123B Instruct 2512 or pi 0.7?+
There is no universal value winner. Compare the input and output prices above with the matched benchmark result for your workload: cheaper tokens can be offset by different quality, token usage, latency, or provider availability.