All evidence cases
Terminal-Bench 2.1terminalProtocol Buffers

Train a CPU-only Caffe model on CIFAR-10

Build BVLC Caffe 1.0.0 for CPU execution and train CIFAR-10 for exactly 500 iterations while meeting the stated accuracy bounds.

caffe-cifar-10

Normalized profile

What this task asks for

Build BVLC Caffe 1.0.0 for CPU execution and train CIFAR-10 for exactly 500 iterations while meeting the stated accuracy bounds. Build Caffe for CPU-only execution. Train for exactly 500 iterations. Evaluate over 100 test iterations. Achieve test accuracy above 45% and within five percentage points of train accuracy.

  • systems setup
  • model training
  • test and validation
  • model training
  • build system
  • solver configuration
  • BVLC Caffe 1.0.0
  • CIFAR-10
  • convolutional neural networks
  • /app/caffe
  • training_output.txt
  • cifar10_quick_iter_500.caffemodel
  • strict behavioral compatibility or exact-output validation
  • quantitative performance, accuracy, or resource constraints
  • specialized algorithmic or domain reasoning
  • changes span multiple components or interfaces

Published evidence

Model × harness outcomes

Counts are tied to these exact configurations. Harnesses are not treated as equivalent across benchmarks.

Models
13
Configurations
20
Trials
100

Showing 20 of 20 configurations

ModelHarnessEffortResultPasses
claude-fable-5anthropic · 2026-06-07claude-code2.1.167xhighAll passed5/5
claude-fable-5anthropic · 2026-06-05terminus-22.0.0highMixed3 errors2/5
claude-opus-4-7anthropic · 2026-05-01claude-code2.1.123maxMixed2 errors3/5
claude-opus-4-7anthropic · 2026-05-01terminus-22.0.0maxMixed2 errors3/5
claude-opus-4-8anthropic · 2026-07-09claude-code2.1.205highAll passed5/5
claude-sonnet-5anthropic · 2026-07-09claude-code2.1.205highMixed1 errors4/5
gemini-3-pro-previewgemini · 2026-05-01gemini-cli0.40.0highMixed2/5
gemini-3-pro-previewgemini · 2026-05-01terminus-22.0.0highMixed1 errors4/5
gemini-3.1-pro-previewgemini · 2026-05-05gemini-cli0.40.0highMixed1 errors3/5
gemini-3.1-pro-previewgemini · 2026-05-05terminus-22.0.0highMixed1 errors3/5
glm-5.1zai · 2026-05-01claude-code2.1.123maxAll passed5/5
gpt-5.5openai · 2026-05-01codex0.125.0xhighAll passed5/5
gpt-5.5openai · 2026-05-01terminus-22.0.0xhighMixed3/5
gpt-5.6-lunaopenai · 2026-07-10codex0.144.0maxAll passed5/5
gpt-5.6-lunaopenai · 2026-07-11codex0.144.1maxAll passed5/5
gpt-5.6-solopenai · 2026-07-10codex0.144.0maxMixed3/5
gpt-5.6-terraopenai · 2026-07-10codex0.144.0maxMixed2 disqualified3/5
gpt-5.6-terraopenai · 2026-07-11codex0.144.1maxMixed4/5
grok-4.5cursor · 2026-07-09cursor-cli2026.07.08-0c04a8adefaultAll passed5/5
muse-spark-1.1openai · 2026-07-09mini-swe-agent2.4.5xhighMixed1 errors4/5