Exaone 4.0 1.2B
Jul 2025
Exaone 4.0 1.2BCompare inon
Benchmarks of Exaone 4.0 1.2B13
BenchmarkByScoreRank
Overall1
Knowledge4
Science2
Coding1
Documents1
Legacy · saturated or no longer reported4
No benchmark matches
—per task?
self-host onlyprice
—output speed?
66Kcontext
Rankingin non-thinking mode?
Alternatives?
ModelScoreCostPer taskWhat changes
BetterGPT OSS 20BhighOpen×3.7—$0.011Memory (4-bit): 0.6 GB → 13 GB
OpenQwen3 4B 2507 (Reasoning)×3.4—$0.40 / $0.80Memory (4-bit): 0.6 GB → 2 GB
BetterMinistral 3 8B 2512Open+32%—$0.011Memory (4-bit): 0.6 GB → 4.5 GB
BetterNemotron 3 Nano 30B A3BthinkingOpen×3.8—$0.017Memory (4-bit): 0.6 GB → 19 GBBetterMinistral 3 14B 2512non-thinkingOpen+25%—$0.019Memory (4-bit): 0.6 GB → 7 GBModelScoreCostPer taskWhat changes
Nothing here yet.
ModelScoreCostPer taskWhat changes
GPT OSS 20BhighOpen×3.7—$0.011Memory (4-bit): 0.6 GB → 13 GBMinistral 3 8B 2512Open+32%—$0.011Memory (4-bit): 0.6 GB → 4.5 GBNemotron 3 Nano 30B A3BthinkingOpen×3.8—$0.017Memory (4-bit): 0.6 GB → 19 GBMinistral 3 14B 2512non-thinkingOpen+25%—$0.019Memory (4-bit): 0.6 GB → 7 GB
DeepSeek V3non-thinkingOpen+8%—$0.02Memory (4-bit): 0.6 GB → 411 GBModelScoreCostPer taskWhat changes
Qwen3 4B 2507 (Reasoning)Open×3.4—$0.40 / $0.80Memory (4-bit): 0.6 GB → 2 GBTFalcon H1R 7BthinkingOpen×3.3——Memory (4-bit): 0.6 GB → 3.8 GBNVIDIA Nemotron Nano 12B v2 VLthinkingOpen×3.1—$0.20 / $0.60Memory (4-bit): 0.6 GB → 6.6 GB
Olmo 3 7B ThinkOpen×2.9——Memory (4-bit): 0.6 GB → 3.6 GBNVIDIA Nemotron Nano 9B V2thinkingOpen×2.9—$0.04 / $0.16Memory (4-bit): 0.6 GB → 4.4 GBRun it yourself
estimateFits on✓ 8 GB (phone, laptop) · full quality✓ 16 GB · full quality✓ 24 GB GPU · full quality✓ 48 GB GPU · full quality✓ 80 GB GPU · full quality
Licenceexaonecommercial use with conditions
Benchmarks
all categoriesChat & Knowledge
Humanity's Last Exam5.7#258
AA Omniscience Index-82.2#320
CritPt (physics)0#182
AA Intelligence Index 4.3.2≈5.2#315
GPQA Diamond42.4#276
AIME 2025 · AA archived24#112
Coding
LiveCodeBench · AA archived29.3#124
Agents & Tool Use
τ²-bench · AA20.5#193
Writing & Language
IFBench (instructions)25.3#231
FAQ
How much does Exaone 4.0 1.2B cost?›
There is no per-unit API price: self-host only.
Is Exaone 4.0 1.2B free to use?›
There is no free API. The weights are open, so you can run it yourself.
How good is Exaone 4.0 1.2B, and what is it best at?›
#256 of 336 in Chat & Knowledge, #210 of 239 in Coding, #191 of 242 in Agents & Tool Use, #228 of 245 in Writing & Language.
Is Exaone 4.0 1.2B better than Qwen3 4B 2507 Instruct?›
On AIME 2025 · AA archived, Exaone 4.0 1.2B scores 50.3 vs 52.3 — lower.
What is the context window of Exaone 4.0 1.2B?›
66K tokens.
When was Exaone 4.0 1.2B released? Is there a newer version?›
Released 2025-07-15. It is the current version in its line.
Is Exaone 4.0 1.2B open source?›
Yes, the weights are open (exaone licence).
Can I run Exaone 4.0 1.2B locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can Exaone 4.0 1.2B read images?›
No, it accepts text only.
What is Exaone 4.0 1.2B?›
Exaone 4.0 1.2B is an open-weights language model from LG AI Research, released Jul 2025.
What are the best alternatives to Exaone 4.0 1.2B?›
GPT OSS 20B (better), Qwen3 4B 2507 (Reasoning) (open weights).