Magistral
MagistralCompare inon
Magistral measured on 4 of 104 Chat & Knowledge benchmarks4
BenchmarkByScoreRank
No benchmark matches
—per task?
price not verifiedprice
—output speed?
—context
Alternatives?
ModelScoreCostPer taskWhat changes
BetterGPT-6 Lunalow+4.6—$0.005Function calling only with none · Chat Completions+3
OpenGLM 5.3 Flashmax+65.8—$0.28Defaults to max · Self-hosted+4BetterGPT-5.6 Lunalow+5.1—$0.01
BetterClaude Haiku 5.5low+2.8—$0.024Remove sampling parameters+5BetterGPT-6.1 Sollow+76.7—$0.13Remove unsupported parameters+5ModelScoreCostPer taskWhat changes
Nothing here yet.
ModelScoreCostPer taskWhat changes
GPT-6 Lunalow+4.6—$0.005Function calling only with none · Chat Completions+3GPT-5.6 Lunalow+5.1—$0.01Claude Haiku 5.5low+2.8—$0.024Remove sampling parameters+5GPT-6.1 Sollow+76.7—$0.13Remove unsupported parameters+5GPT-6 Sollow+31.5—$0.13ModelScoreCostPer taskWhat changes
GLM 5.3 FlashmaxOpen+65.8—$0.28Defaults to max · Self-hosted+4
DeepSeek V4 Pro 0813maxOpen+61.3—$0.67Pass back reasoning_content · Chat Completions+4
Kimi K3maxOpen+60.4—$2.00Sampling parameters are fixed+3
Qwen3.8 27BxhighOpen+42.4—$1.01max_tokens range · QwenCloud+4TInkling SmallxhighOpen+40.1—$0.45 / $1.20No strict JSONModelScoreCostPer taskWhat changes
Magistral Medium 1mediumsame——Magistral Small 1unstatedOpensame—$0.70 / $1.16Mistral Large 4thinkingn/a—$1.13Infinite whitespace stream+4Mistral Medium 3.5highOpenn/a—$0.50Mistral Small 4thinkingOpenn/a—$0.015Ministral 3 14B 2512non-thinkingOpenn/a—$0.019Ministral 3 3B 2512Openn/a—$0.008Ministral 3 8B 2512Openn/a—$0.011All Mistral models →Benchmarks
in Chat & KnowledgeARC-AGI-1 · semi-private6.1#57
FAQ
How much does Magistral cost?›
There is no per-unit API price: price not verified.
How good is Magistral, and what is it best at?›
#59 of 66 in Chat & Knowledge.
Is Magistral better than Magistral Small 1?›
On ARC-AGI-2 · semi-private, Magistral scores 0 vs 0 — about the same.
What is Magistral?›
Magistral is a language model from Mistral.
What are the best alternatives to Magistral?›
GPT-6 Luna (better), GLM 5.3 Flash (open weights).