Unverified50% confidenceBenchmarkExact time
In the Needle In A Haystack test, Mistral Large 3 achieved 99.8% retrieval accuracy at any depth quadrant of the half-million-token window.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/12/2026
First Seen
Valid until: 1/10/2027
Sources
Related Entities
Related Claims
UnverifiedOn the SWE-bench Verified benchmark, Mistral Large 3 achieved a score of 53.8%.69% similarUnverifiedMistral Large 4 scores 93% on the cybersecurity benchmark Cybench (40 CTF challenges), 82% on vulnerability reproduction and patching tests, and 93.3% on the B3 AI security benchmark.63% similarUnverifiedIndependent testing shows Mistral Large 4 achieved a success rate of 81.7% on the CyberGym E2E end-to-end test, the highest among 18 tested models.63% similarUnverifiedMistral Large 4 scored approximately 62% on the real-world software engineering benchmark DeepSwee.60% similarUnverified通义千问3.6 27B在语义三次全对率和格式正确率上均领先Muse Glimmer59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1008182API
curl https://kongchang.com/api/v1/knowledge/claims/1008182MCP
get_claim(id=1008182)