SShogo Bench

Llama 2-70B

Llama 2-70B from Meta.

Shogo Agent Index

Unranked

Not ranked yet: 0 agentic benchmarks measured, and we need at least 2 plus an overall capability score. It will be ranked as results are published.

Agentic coding
4
estimated
Long-horizon autonomy
9
estimated
Knowledge-work agents
7
estimated
Computer use
0
estimated
Security & context learning
3
estimated

Price

No public API price.

Agent-effective = 20 input tokens per output token, 85% cache hits. Why.

Agentic benchmarks

BenchmarkScoreNormalisedIndex weightHarness / source
SWE-bench Verifiedest.8%predicted 13
Terminal-Bench 2.0est.8%predicted 0
DeepSWEest.6%predicted 0
FrontierSWEest.5%predicted 0
CursorBenchest.4%predicted 4
GSO-Benchest.4%predicted 0
METR Time Horizonsest.10%predicted 16
Vending-Bench 2est.6%predicted 0
PostTrainBenchest.4%predicted 3
APEX-Agentsest.8%predicted 1
GDPvalest.7%predicted 2
Remote Labor Indexest.5%predicted 0
DeepResearch Benchest.5%predicted 32
OSWorld 2.0est.6%predicted 0
OSWorldest.4%predicted 0
Cybenchest.5%predicted 1
CL-benchest.5%predicted 5

Other benchmarks

BenchmarkCategoryScoreSource
DTBenchKnowledge41.6%source
GPQA DiamondReasoning26.3%source
MATH Level 5Math3.3%source
MMLULegacy69.9%source
OTIS Mock AIMEMath0.0%source