INTELLECT 3 measured on 18 of 104 Chat & Knowledge benchmarks18
BenchmarkByScoreRank
Overall1
Knowledge12
Science2
Documents1
Legacy · saturated or no longer reported2
No benchmark matches
—per task?
self-host onlyprice
—output speed?
131Kcontext
Alternatives?
ModelScoreCostPer taskWhat changes
BetterGPT-6 Lunalow+55%—$0.005Function calling only with none · Chat Completions+3
OpenQwen3.8 Flashthinking×2.9—$0.37Memory (4-bit): 64 GB → 108 GB+6BetterGPT-5.6 Lunalow+52%—$0.01
BetterMiMo v2.5thinkingOpen×2.1—$0.019Memory (4-bit): 64 GB → 186 GB
BetterClaude Haiku 5.5low×2.1—$0.024Remove sampling parameters+5ModelScoreCostPer taskWhat changes
Nothing here yet.
ModelScoreCostPer taskWhat changes
GPT-6 Lunalow+55%—$0.005Function calling only with none · Chat Completions+3GPT-5.6 Lunalow+52%—$0.01MiMo v2.5thinkingOpen×2.1—$0.019Memory (4-bit): 64 GB → 186 GBClaude Haiku 5.5low×2.1—$0.024Remove sampling parameters+5
Mercury 2high+31%—$0.039Context: 131K → 128KModelScoreCostPer taskWhat changes
Qwen3.8 FlashthinkingOpen×2.9—$0.37Memory (4-bit): 64 GB → 108 GB+6Qwen3.8 27BxhighOpen×2.6—$1.01max_tokens range · QwenCloud+4TInkling SmallthinkingOpen×2.5—$0.089Memory (4-bit): 64 GB → 160 GB+1
Nemotron 3.5 Super VLOpen×2.4—$0.19Memory (4-bit): 64 GB → 75 GB
MiniMax M2.7thinkingOpen×2.3—$0.30 / $1.20Memory (4-bit): 64 GB → 137 GBRun it yourself
estimateFits on✗ 8 GB (phone, laptop)✗ 16 GB✗ 24 GB GPU✗ 48 GB GPU✓ 80 GB GPU · compressed
LicenceMITfree for commercial use
Benchmarks
in Chat & KnowledgeAA Intelligence Index 4.3.2≈10.6#203
AA Omniscience Index-50.9#214
AA Omniscience · Accuracy19.6#159
Humanity's Last Exam13.1#155
GPQA Diamond76.1#144
AA-LCR (long context) 1.139.7#199
IFBench (instructions)34#193
CritPt (physics)0.3#149
FAQ
How much does INTELLECT 3 cost?›
There is no per-unit API price: self-host only.
Is INTELLECT 3 free to use?›
There is no free API. The weights are open, so you can run it yourself.
How good is INTELLECT 3, and what is it best at?›
#154 of 334 in Chat & Knowledge, #128 of 239 in Coding, #161 of 242 in Agents & Tool Use, #126 of 195 in Writing & Language.
Is INTELLECT 3 better than ERNIE 5.0 Thinking Preview?›
On Humanity's Last Exam, INTELLECT 3 scores 13.1 vs 13.3 — about the same.
What is the context window of INTELLECT 3?›
131K tokens.
When was INTELLECT 3 released? Is there a newer version?›
Released 2025-11-27. It is the current version in its line.
Is INTELLECT 3 open source?›
Yes, the weights are open (MIT licence).
Can I run INTELLECT 3 locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can INTELLECT 3 read images?›
No, it accepts text only.
What is INTELLECT 3?›
INTELLECT 3 is an open-weights language model from Prime Intellect, released Nov 2025.
What are the best alternatives to INTELLECT 3?›
GPT-6 Luna (better), Qwen3.8 Flash (open weights).