Step 5 Preview — Benchmarks

Benchmark scores for Step 5 Preview aggregated from public leaderboards (3 boards), with how it ranks among open models and how it compares with its nearest neighbours. See hardware requirements for what you need to run it.

llmrun Composite Score →

128.0

90% range 124.3–131.8 · from 3 boards

CodingNot enough data yet
AgentsNot enough data yet
MathNot enough data yet
Science134.290% range 122–147
ReasoningNot enough data yet

Step 5 Preview reasoning benchmarks

Multi-step logic and problem-solving — can the model think through a hard problem rather than recall an answer.

CritPt
CritPt (Critical Points) tests graduate-level physics problem-solving on research-style questions spanning mechanics, electromagnetism, quantum mechanics, and more. It was created by physicists to probe whether a model can reason through multi-step physics rather than pattern-match textbook answers.
20.9%
20.9%
median 1.1%open leader 26.6%
Overall best: GPT 5.6 Sol Max 32.3%
96th percentile · #3 of 68 open · #54 of 220 all
#3 of 68
#54 of 220
Lab-reported
Epoch AI · Sep 2026

Step 5 Preview coding benchmarks

Writing, editing, and fixing real code — measured by whether the resulting program actually runs and passes tests.

SciCode
SciCode asks a model to write real scientific-computing code — implementing multi-step numerical methods drawn from physics, chemistry, biology, and math research — and checks the output against reference solutions. It was built by a team of scientists and ML researchers to test coding ability on genuine research problems, not typical software tasks.
58.9%
58.9%
median 39.5%open leader 60.9%
Overall best: Claude Opus 5.5 Max 66.9%
96th percentile · #4 of 68 open · #16 of 208 all
#4 of 68
#16 of 208
Lab-reported
Epoch AI · Sep 2026

Step 5 Preview math benchmarks

Symbolic and quantitative problem-solving, from competition math to multi-step calculation.

ProofBench
ProofBench evaluates a model's ability to write and check rigorous mathematical proofs, rather than just produce a final numeric answer. It probes step-by-step logical correctness, which is a harder and more distinct skill than solving for a number.
42.0%
42.0%
median 18.0%open leader 87.0%
Overall best: Claude Fable 5.1 Max 100.0%
69th percentile · #9 of 27 open · #38 of 82 all
#9 of 27
#38 of 82
Lab-reported
Epoch AI · Sep 2026

Score vs VRAM in the ≤512 GB tier

Step 5 Preview
Other models
Best Score for its VRAM
x: VRAM (GB, log scale) · y: llmrun Score

#6 of 98 that fit ≤512 GB

Each dot is a model at the best quant that fits ≤512 GB. The line joins the models with the best Score for their VRAM; a dot below it is beaten by a model that needs no more memory.

Full ≤512 GB ranking →

Step 5 Preview vs its neighbours

Same boards for the nearest open models per llmrun Score. The highest score in each row is marked.

Board	Step 5 Previewthis model	GLM 5.3nearest by ScoreQ4_K_M · 454 GB	DeepSeek V4.1 Flashnearest by ScoreQ4_K_M · 458 GB	Muse Spark Open 816B A42Bnearest by ScoreQ4_K_M · 552 GB
llmrun Score	128.0	130.1	129.4	125.6
CritPt	20.9%	19.1%	14.3%	14.9%
SciCode	58.9%	59.0%	51.8%	58.5%
ProofBench	42.0%	49.0%	54.0%	—
Compare head to head		Compare	Compare	Compare

Scores aggregated from public benchmark sources (Epoch AI: 3 boards) — each linked from the benchmark pages. llmrun does not run these benchmarks. Percentiles and ranks are among open models in the llmrun catalogue; the median and leader marks refer to the same field.