Mistral Large 3 coding benchmark
Mistral AI · API model Mistral-Large-3 · Results published 25 Oct 2026 · Benchmark v4
How Mistral Large 3 performed
Mistral Large 3 by Mistral AI took the PomBench AI coding benchmark: every model gets the same prompt and writes self-contained browser code for a twelve-second scene: a Pomeranian runs across a backyard, jumps a puddle, enters its doghouse, retrieves a bone and turns toward the viewer. The video above is an independent offline browser recording of the code Mistral Large 3 returned, captured at 960 × 540 and 30 fps, with no organizer edits to the code.
In this Mistral Large 3 coding test the model produced 5,448 output tokens, including provider-reported reasoning (rank 31 of 46 by output tokens), and the known API cost was $0.0092 (rank 11 of 45, cheapest first). Is Mistral Large 3 good at coding? One task cannot answer that on its own, so this LLM coding benchmark lets you watch the result instead of reading a score: check the gait, the puddle jump and reflection, the doorway occlusion and whether the bone stays in the dog's mouth.
Then compare Mistral Large 3 vs Mistral Medium 3.5 and Codestral 2501 below, or open the AI model comparison to play up to four models on one synchronized timeline.
Mistral Large 3 vs Mistral Medium 3.5, Codestral 2501 and Claude Haiku 4.5
- Mistral Large 3 vs Mistral Medium 3.5: 5,448 vs 3,499 output tokens; known API cost $0.0092 vs $0.0294. Watch side by side
- Mistral Large 3 vs Codestral 2501: 5,448 vs 1,440 output tokens; known API cost $0.0092 vs $0.0019. Watch side by side
- Mistral Large 3 vs Claude Haiku 4.5: 5,448 vs 7,233 output tokens; known API cost $0.0092 vs $0.0384. Watch side by side
Mistral Large 3 code
Mistral Large 3 wrote its own native browser animation code from the same prompt as every other model. One execution repair is available for empty, truncated or unexecutable code; the final response is kept, not a best-of selection.