‹ PublicAI Index
The LLM benchmark aggregator.
Beam 1.0
Reflection
Strongest in Web research (#25 of 49, on 1 of its 2 boards), weakest in Agents (#186 of 289). Above par in 4 of 17 scopes. Among the models it meets almost everywhere, it finishes behind GPT-6 Astra and GPT-6.1 Sol and ahead of DeepSeek V3 and Mistral Large 3.
§ 1 · Profile
What it is good at
Bars run from 50 — the average of the models each source lists — so right of the line is above par. The middle column is the gap to whoever leads that scope.
Coding53.4−13#51/1762/6
Agentic coding53.6−13.4#49/1692/5
Reasoning43−21.9#151/1903/5
Reasoning45.2−21.5#87/1442/3
Mathematics39.2−24.9#121/1501/2
Core abilities46−19.2#152/2252/4
Data analysis42.4−23.7#37/481/1
General intelligence47−21.8#130/2142/3
Professional42.4−20.7#165/1882/2
Biology research26.1−42.1#28/281/1
Document design42−25.3#37/481/1
Legal43.7−22.2#125/1641/1
Agents46.1−19.8#186/2893/6
Web research51.7−10.9#25/491/2
Long-horizon work37−37#40/481/1
Knowledge work54.8−17#60/1982/2
Computer use33.2−34.5#77/811/2
§ 2 · Head to head
What it beats, and what beats it
The same models turn up scope after scope. Counted once: where both were placed, who finished higher. Bars run right for Beam 1.0, left for the other.
§ 3 · Sources
Where the numbers come from
6 publications, 17 figures. Every one links to the page it was read from.
Artificial Analysis Intelligence Index 30
GDPval-AA 1177
AA-Briefcase 976
ARC-AGI-2 23.5%
LiveBench 55.5LiveBench · Agentic 17.3LiveBench · Reasoning 80.2LiveBench · Coding 83.9LiveBench · Browser Use 0LiveBench · Data Analysis 66.4LiveBench · Research 72.8LiveBench · Design 67.9
Vals · Legal Research Bench 28.85%Vals · Harvey Legal Agent Benchmark 4.58%Vals · BioMysteryBench 42.59%Vals · Vibe Code Bench 66.44% (OpenHands)Vals · ProofBench 18%
Badge
[](https://publicai.io/model-index/m/beam-1-0)