Skip to main content
Radar

Five or fewer confirmed AI changes, with original sources, on mornings when something changed.A free source-linked morning brief.

See the free Radar Brief

InferenceBench

A benchmark for open-ended LLM inference optimization by AI agents. Agents receive a base model, one H100, and a fixed time budget to build a valid OpenAI-compatible inference server that improves serving speed.

Data verified 23 confirmed releases in the last 30 daysSee the free Radar Brief

How BenchLM shows InferenceBench

BenchLM mirrors the public InferenceBench agent leaderboard captured on May 20, 2026. The source evaluates 36 frontier agent configurations over 4 inference-serving scenarios with a 2 h budget on 1 NVIDIA H100 80 GB.

InferenceBench is display only on BenchLM. It is a useful agentic systems-engineering signal, but the rows combine model capability, agent scaffold, inference framework choices, hardware, and final-server validity, so BenchLM keeps it separate from weighted model-only rankings.

Snapshot

36 agent rows4 scenarios180 runs2 hDisplay only

Aggregate speedup on InferenceBench — May 20, 2026

We mirror the published aggregate speedup view for InferenceBench. Claude Opus 5 leads the public snapshot at 8.90x, followed by Claude Fable 5 (8.74x) and Claude Opus 4.7 (8.53x). We do not use these results to rank models overall.

36 modelsAgenticCurrentDisplay onlyUpdated May 20, 2026

Aggregate speedup table (36 models)

Score
1
Claude Opus 5Anthropic · ClosedClaude Code · v2.1.119 · strict prompt
8.90x
2
Claude Fable 5Anthropic · ClosedClaude Code · v2.1.175 · strict prompt
8.74x
3
Claude Opus 4.7Anthropic · ClosedClaude Code · v2.1.175
8.53x
4
Claude Opus 4.8Anthropic · ClosedClaude Code · v2.1.175
7.60x
5
Claude Fable 5Anthropic · ClosedClaude Code · v2.1.175 · strict prompt
7.52x
6
GPT-5.6 SolOpenAI · ClosedCodex CLI · strict prompt
7.34x
7
Claude Opus 4.8Anthropic · ClosedClaude Code
7.34x
8
GLM-5.2Z.AI · Open weightClaude Code · v2.1.119 · strict prompt
7.00x
9
Grok 4.6xAI · ClosedGrok Build · strict prompt
6.65x
10
Claude Sonnet 5Anthropic · ClosedClaude Code · v2.1.119 · strict prompt
6.43x
11
Kimi K2.7 CodeMoonshot AI · Open weightOpenCode · strict prompt
6.27x
12
GPT-5.4OpenAI · ClosedCodex CLI
6.16x
13
Kimi K3Moonshot AI · ClosedOpenCode · strict prompt
5.70x
14
Claude Sonnet 4.6Anthropic · ClosedClaude Code
5.56x
15
GPT-5.3 CodexOpenAI · ClosedCodex CLI
5.49x
16
GPT-5.5OpenAI · ClosedCodex CLI
5.45x
17
GLM-5.3Z.AI · Open weightClaude Code · v2.1.119 · strict prompt
4.98x
18
Gemini 3.1 ProGoogle · ClosedOpenCode
4.92x
19
Kimi K2.6Moonshot AI · Open weightOpenCode
4.51x
20
Ox Alpha (stealth)UnknownOpenCode · strict prompt
4.49x
21
Claude Opus 4.6Anthropic · ClosedClaude Code
4.38x
22
GPT-5.2OpenAI · ClosedCodex CLI
4.28x
23
GPT-5.5OpenAI · ClosedCodex CLI
4.22x
24
Gemini 3.5 FlashGoogle · ClosedOpenCode
4.16x
25
Claude Opus 4.5Anthropic · ClosedClaude Code
3.76x
26
Grok 4.5xAI · ClosedGrok Build · strict prompt
3.70x
27
GPT-5.1-Codex-MaxOpenAI · ClosedCodex CLI
3.59x
28
Grok 4.5xAI · ClosedOpenCode · strict prompt
3.42x
29
GLM-5Z.AI · Open weightOpenCode
3.22x
30
Claude Sonnet 4.5Anthropic · ClosedClaude Code
3.18x
31
Claude Fable 5Anthropic · ClosedClaude Code · v2.1.175
3.16x
32
Claude Haiku 4.5Anthropic · ClosedClaude Code
2.78x
33
GPT-5.3 CodexOpenAI · ClosedCodex CLI
2.32x
34
Claude Opus 4.7Anthropic · ClosedClaude Code · v2.1.114
2.25x
35
Claude Fable 5Anthropic · ClosedClaude Code · v2.1.175
2.15x
36
GPT-5.2-CodexOpenAI · ClosedCodex CLI
1.98x

The published InferenceBench snapshot places Claude Opus 5 first at 8.90x. The third row is 0.37 points behind. The broader top-10 range is 2.47 points, so many of the published results sit in a relatively narrow band.

36 models have been evaluated on InferenceBench. The benchmark falls in the Agentic category. This category carries a 22% weight in BenchLM.ai's overall scoring system. InferenceBench is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.

About InferenceBench

Year

2026

Tasks

4 inference-serving optimization scenarios

Format

Two-hour autonomous CLI agent run

Difficulty

Open-ended ML systems engineering

BenchLM mirrors the public InferenceBench agent leaderboard as a display-only agentic systems-engineering benchmark. The primary score is aggregate geometric-mean speedup over a PyTorch baseline across prefill latency, decode latency, throughput, and all-in-one serving scenarios.

BenchLM freshness & provenance

Version

InferenceBench 2026

Refresh cadence

Quarterly

Staleness state

Current

Question availability

Public benchmark harness and aggregate leaderboard

CurrentDisplay only

BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.

FAQ

What does InferenceBench measure?

A benchmark for open-ended LLM inference optimization by AI agents. Agents receive a base model, one H100, and a fixed time budget to build a valid OpenAI-compatible inference server that improves serving speed.

Which model leads the published InferenceBench snapshot?

Claude Opus 5 currently leads the published InferenceBench snapshot with 8.90x aggregate speedup. BenchLM shows this benchmark for display only and does not use it in overall rankings.

How many models are evaluated on InferenceBench?

The May 20, 2026 snapshot contains 36 AI models.

Last updated: May 20, 2026 · mirrored from the public benchmark leaderboard

Know when it’s worth switching models

The model to choose, the cheaper alternative, and the release we would wait on.

Read a sample issue

Join 2,000+ readers.

One email each week. Unsubscribe anytime.