llmboard.aiAI model intelligence
Home

Model Rankings

OverallOpen ModelsAgentCodingReasoningMathKnowledgeInstruction FollowingTextVision
Image GenerationImage Editing
Video GenerationImage to VideoVideo Editing
Text to SpeechSpeech to Text
Embeddings

Efficiency

Chat Token PricingImage PricingVideo PricingAudio Pricing
Chat Speed & LatencyProvider Reliability

Benchmarks

GPQAMMLU-ProAIME 2025SWE-Bench VerifiedMMLUHumanity's Last ExamLiveCodeBenchMATHHumanEvalMMMU-Pro
All Benchmarks

Tools

Model Directory

Scoring & Data

Scoring & Data
1464 models816 benchmarks

Leaderboard Center

Overall RankingCodingCore BenchmarksPrice & ValueRuntime Performance

Modalities

All ModelsImage GenerationImage EditingVideo GenerationImage-to-VideoVideo EditingText-to-SpeechSpeech-to-TextEmbeddings

Data & Methods

Scoring MethodAll Benchmarks
llmboard.aiCopyright 2026 llmboard.ai

agents benchmark

Ramp SWE-Bench v2 (Bash) Leaderboard

Ramp SWE-Bench v2 (Bash) is a private coding benchmark with 500 backend engineering tasks based on pull requests merged into production after engineer review.

Updated Nov 22, 2026

Models22
Model coverage22
MetricScore
EvidenceB

On this page

  • Ranking
  • Highlights
  • Distribution
  • Top models
  • About
  • FAQ

Ramp SWE-Bench v2 (Bash) Ranking

Higher score ranks better on this benchmark.

22 rows
Columns

Show columns

Sort by
Rank
Model
Score
Percentile
Participants
Evidence
Evaluated
Rank01ModelOPGPT-6.2 AstraOpenAIScore75.00%Percentile100.00%Participants22EvidenceBEvaluatedNov 21, 2026
Rank02ModelGOGemini 4 CarbonGoogleScore74.00%Percentile95.24%Participants22EvidenceBEvaluatedNov 21, 2026
Rank03ModelANClaude Fable 5.2AnthropicScore73.00%Percentile90.48%Participants22EvidenceBEvaluatedNov 21, 2026
Rank04ModelOPGPT-6.2 SolOpenAIScore72.40%Percentile85.71%Participants22EvidenceBEvaluatedNov 21, 2026
Rank05ModelANClaude Opus 5.5AnthropicScore71.20%Percentile80.95%Participants22EvidenceBEvaluatedNov 21, 2026
Rank06ModelGOGemini 4 HydrogenGoogleScore70.80%Percentile76.19%Participants22EvidenceBEvaluatedNov 21, 2026
Rank07ModelANClaude Fable 5.1AnthropicScore63.80%Percentile71.43%Participants22EvidenceBEvaluatedNov 21, 2026
Rank08ModelOPGPT-6.1 TerraOpenAIScore60.60%Percentile66.67%Participants22EvidenceBEvaluatedNov 21, 2026
Rank09ModelMEMuse WonderMetaScore57.80%Percentile61.90%Participants22EvidenceBEvaluatedNov 21, 2026
Rank10ModelANClaude Sonnet 5.5AnthropicScore57.60%Percentile57.14%Participants22EvidenceBEvaluatedNov 21, 2026
Rank11ModelXAGrok 4.7xAIScore55.40%Percentile52.38%Participants22EvidenceBEvaluatedNov 21, 2026
Rank12ModelGOGemini 4 ArgonGoogleScore54.80%Percentile47.62%Participants22EvidenceBEvaluatedNov 21, 2026
Rank13ModelOPGPT-6 AstraOpenAIScore54.60%Percentile42.86%Participants22EvidenceBEvaluatedNov 21, 2026
Rank14ModelMEMuse EmberMetaScore54.40%Percentile38.10%Participants22EvidenceBEvaluatedNov 21, 2026
Rank15ModelDEDeepSeek-V4.1-Flash-1017DeepSeekScore53.40%Percentile33.33%Participants22EvidenceBEvaluatedNov 21, 2026
Rank16ModelANClaude Haiku 5.5AnthropicScore52.80%Percentile28.57%Participants22EvidenceBEvaluatedNov 21, 2026
Rank17ModelOPGPT-6.1 SolOpenAIScore52.00%Percentile23.81%Participants22EvidenceBEvaluatedNov 21, 2026
Rank18ModelGOGemini 3.8 FlashGoogleScore47.00%Percentile19.05%Participants22EvidenceBEvaluatedNov 21, 2026
Rank19ModelMAKimi K3.1 ProMoonshot AIScore45.40%Percentile14.29%Participants22EvidenceBEvaluatedNov 21, 2026
Rank20ModelOPGPT-6 LunaOpenAIScore42.60%Percentile9.52%Participants22EvidenceBEvaluatedNov 21, 2026
Rank21ModelRABeam 1.0Reflection AIScore34.00%Percentile4.76%Participants22EvidenceBEvaluatedNov 21, 2026
Rank22ModelNVNemotron 3.5 Super VL (120B A12B)NVIDIAScore12.00%Percentile0.00%Participants22EvidenceBEvaluatedNov 21, 2026

Ramp SWE-Bench v2 (Bash) Highlights

The leading models and scores on this benchmark.

Ramp SWE-Bench v2 (Bash) Score Distribution

A closer view of the leading scores on this benchmark.

Ramp SWE-Bench v2 (Bash)

The Top AI Models for Ramp SWE-Bench v2 (Bash)

The first five results on this benchmark, with official price and output speed added where the model identity can be matched.

What is Ramp SWE-Bench v2 (Bash)?

What Ramp SWE-Bench v2 (Bash) measures and how its scores work.

Each model receives a one-shot prompt in the same sandboxed copy of Ramp's codebase and produces a review-ready patch using the bash-only mini-swe-agent harness.

It measures Score as pass@1 ratio at high reasoning effort with web tools disabled; a task passes when an LLM judge finds no material difference where the engineer-reviewed merged solution is better.

Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.

Family
Ramp SWE-Bench v2 (Bash)
Modality
text
Primary category
agents
Score direction
higher
LLMBoard eligible
No
Evaluation key
overall

LLM Stats. Benchmark scores retain their original unit. Overall score eligibility is shown separately.

FAQ

Common questions about Ramp SWE-Bench v2 (Bash).

Which model scores highest on Ramp SWE-Bench v2 (Bash)?

GPT-6.2 Astra is currently ranked first with 75.00%.

What are the top three models on Ramp SWE-Bench v2 (Bash)?

The current leaders are GPT-6.2 Astra (75.00%), Gemini 4 Carbon (74.00%), and Claude Fable 5.2 (73.00%).

Which Ramp SWE-Bench v2 (Bash) model has the lowest official input price?

Muse Ember has the lowest matched official input price at $0.10 input / $0.50 output per 1M tokens.

Which models are fastest among Ramp SWE-Bench v2 (Bash) results?

The fastest matched records are Nemotron 3.5 Super VL (120B A12B) (96.29 tok/s via DeepInfra), Muse Ember (63.51 tok/s via Meta Model API), and Claude Sonnet 5.5 (42.00 tok/s via Anthropic).

Does the highest score result prove overall model quality?

No. This benchmark measures one defined capability or task. The overall LLMBoard score uses a separate aggregation across eligible benchmark evidence.

What does Ramp SWE-Bench v2 (Bash) measure?

It measures Score as pass@1 ratio at high reasoning effort with web tools disabled; a task passes when an LLM judge finds no material difference where the engineer-reviewed merged solution is better.

Is a higher score better?

Yes. Higher values rank better for this benchmark.

How many models are compared?

22 model results are currently shown.

Does this benchmark affect the overall score?

No. This benchmark is shown for reference but does not contribute to the overall score.

Rank #1GPT-6.2 Astra75.00%
Rank #2Gemini 4 Carbon74.00%
Rank #3Claude Fable 5.273.00%
Rank #4GPT-6.2 Sol72.40%

Ranking basisThis ramp swe-bench v2 (bash) AI model leaderboard uses descending score in the benchmark original unit. The leaderboard ranking keeps matched price and speed data separate from benchmark evidence.

  1. 01
    OP
    GPT-6.2 AstraOpenAI
    Score
    75.00%
    Price
    $8.0 input / $40 output per 1M tokens

    Strengths

    • Ranks #1 of 22 compared models
    • 100th percentile on this benchmark
    • B evidence result

    Considerations

    • This result measures Ramp SWE-Bench v2 (Bash), not total model capability
  2. 02
    GO
    Gemini 4 CarbonGoogle
    Score
    74.00%
    Price
    $2.0 input / $10 output per 1M tokens
    Speed
    Up to 11.31 tok/s via Google

    Strengths

    • Ranks #2 of 22 compared models
    • 95th percentile on this benchmark
    • B evidence result

    Considerations

    • This result measures Ramp SWE-Bench v2 (Bash), not total model capability
  3. 03
    AN
    Anthropic
    Score
    73.00%
    Price
    $8.0 input / $40 output per 1M tokens

    Strengths

    • Ranks #3 of 22 compared models
    • 90th percentile on this benchmark
    • B evidence result

    Considerations

    • This result measures Ramp SWE-Bench v2 (Bash), not total model capability
  4. 04
    OP
    OpenAI
    Score
    72.40%
    Price
    $2.0 input / $10 output per 1M tokens

    Strengths

    • Ranks #4 of 22 compared models
    • 86th percentile on this benchmark
    • B evidence result

    Considerations

    • This result measures Ramp SWE-Bench v2 (Bash), not total model capability
  5. 05
    AN
    Anthropic
    Score
    71.20%
    Price
    $4.0 input / $20 output per 1M tokens

    Strengths

    • Ranks #5 of 22 compared models
    • 81th percentile on this benchmark
    • B evidence result

    Considerations

    • This result measures Ramp SWE-Bench v2 (Bash), not total model capability

Selection summary

Best AI Models for Ramp SWE-Bench v2 (Bash)

GPT-6.2 Astra currently leads Ramp SWE-Bench v2 (Bash) with 75.00%. It is the top model on this specific benchmark, while the best LLM for the broader task should also be checked against other benchmarks, price, and runtime.

Use this leaderboard with the supporting benchmark results and coverage details above. A leaderboard position summarizes the selected ranking signal; it does not replace workload-specific testing.

Claude Fable 5.2
GPT-6.2 Sol
Claude Opus 5.5
Benchmark rank #1GPT-6.2 Astra75.00% · $8.0 input / $40 output per 1M tokens
Benchmark rank #2Gemini 4 Carbon74.00% · $2.0 input / $10 output per 1M tokens
Benchmark rank #3Claude Fable 5.273.00% · $8.0 input / $40 output per 1M tokens