Mixtral 8x22B Instruct
Mixtral 8x22B InstructRankingin non-thinking mode?
Strong at
Alternatives?
CheaperMuse Spark 1.3max×8.4-33%$1.60Only auto tool_choice is supported+5
BetterGPT-6 Lunalow×3.7-93%$0.005Function calling only with none · Chat Completions+3
OpenGLM 5.3 Flashmax×7.3-92%$0.28Memory (4-bit): 84 GB → 193 GB+5
CheaperMiMo v2.6 ProthinkingOpen×8.1-82%$0.13Memory (4-bit): 84 GB → 615 GB+4
CheaperDeepSeek V4.1 Flash 1017maxOpen×7.9-83%$0.26Memory (4-bit): 84 GB → 458 GB+5Muse Spark 1.3max×8.4-33%$1.60Only auto tool_choice is supported+5MiMo v2.6 ProthinkingOpen×8.1-82%$0.13Memory (4-bit): 84 GB → 615 GB+4DeepSeek V4.1 Flash 1017maxOpen×7.9-83%$0.26Memory (4-bit): 84 GB → 458 GB+5GLM 5.3maxOpen×7.8-28%$2.01Memory (4-bit): 84 GB → 452 GB+4
Step 5 PreviewthinkingOpen×7.6-52%$1.03Memory (4-bit): 84 GB → 363 GB+3GPT-6 Lunalow×3.7-93%$0.005Function calling only with none · Chat Completions+3
Granite 4.2 3BthinkingOpen+58%-98%$0.006
Gemma 4 E4BthinkingOpen+55%-99%$0.008GPT-5.6 Lunalow×3.7-85%$0.01GPT OSS 20BhighOpen+56%-99%$0.011GLM 5.3 FlashmaxOpen×7.3-92%$0.28Memory (4-bit): 84 GB → 193 GB+5
Qwen3.8 FlashthinkingOpen×6.9-92%$0.37Memory (4-bit): 84 GB → 108 GB+6MiMo v2.6 FlashthinkingOpen×6.6-94%$0.062Memory (4-bit): 84 GB → 186 GB+4Qwen3.8 27BxhighOpen×5.9-68%$1.01max_tokens range · QwenCloud+4MMotif 3Open×5.8-71%$0.50 / $2Memory (4-bit): 84 GB → 189 GBMistral Large 4thinking×6.7—$1.13Infinite whitespace stream+4Mistral Medium 3.5highOpen×2.5—$0.50Mistral Small 4thinkingOpen+96%—$0.015Mistral Large 3non-thinkingOpen+61%—$0.031Memory (4-bit): not verifiedMagistral Medium 1thinking+58%——Context: 66K → 40KDevstral Mediumnon-thinkingOpen+57%—$0.40 / $2Memory (4-bit): not verifiedMagistral Small 1thinkingOpen+43%—$0.70 / $1.16Mistral Small 3non-thinkingOpensame—$0.05 / $0.08Context: 66K → 32KAll Mistral models →Run it yourself
estimateBenchmarks
all categoriesFAQ
How much does Mixtral 8x22B Instruct cost?›
$2 per 1M input tokens and $6 per 1M output tokens; cached input $0.20 per 1M.
Is Mixtral 8x22B Instruct free to use?›
There is no free API; the cheapest paid price is $2 per 1M input tokens. The weights are open, so you can run it yourself.
How good is Mixtral 8x22B Instruct, and what is it best at?›
#308 of 334 in Chat & Knowledge, #140 of 152 in Coding, #179 of 195 in Writing & Language.
Is Mixtral 8x22B Instruct better than Mistral Large?›
On AA Intelligence Index 4.3.2, Mixtral 8x22B Instruct scores 5.7 vs 5.8 — about the same.
How can I use Mixtral 8x22B Instruct?›
Through the API of Mistral.
What is the context window of Mixtral 8x22B Instruct?›
66K tokens, with up to 52K output tokens.
When was Mixtral 8x22B Instruct released? Is there a newer version?›
Released 2024-04-17. It is the current version in its line.
Is Mixtral 8x22B Instruct open source?›
Yes, the weights are open (Apache-2.0 licence).
Can I run Mixtral 8x22B Instruct locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can Mixtral 8x22B Instruct read images?›
It accepts text and PDF files as input.
What is Mixtral 8x22B Instruct?›
Mixtral 8x22B Instruct is an open-weights language model from Mistral, released Apr 2024.
What are the best alternatives to Mixtral 8x22B Instruct?›
Muse Spark 1.3 (cheaper), GPT-6 Luna (better), GLM 5.3 Flash (open weights).