Benchmarks
Independent, reproducible measurements of the models and workflows behind great customer support. Every score links to the configuration, cost, and telemetry behind it.
Fetch benchmark results and metadata through the API.
Showing all benchmark categories
Agents & tools
Media
Generated images, clips, and speech, graded against the request and priced per output.
Image
Image prompts built to fail: depth, direction, counting, and text in the frame.
44 models
Samples
Value
$0.010
Meta: Muse Image
Speed
4s
Google: Nano Banana 2 Lite
Video
Six seconds of video, held to one duration and resolution across every model.
24 models
Samples
Value
$0.24
ByteDance: Seedance 2.0 Mini
Speed
56s
MiniMax: H3 Max
Reasoning
Search
BrowseComp
Hard-to-locate facts on the live web, scored on persistent multi-step research.
4 modelslast run Aug 18, 2026
Quality
89.0%
Claude Opus 5 · high
Value
$0.99
Claude Opus 5 · high
Speed
1.9m
Claude Opus 5 · high
DeepSearchQA
Questions whose answers are lists, scored for exhaustive retrieval with no padding.
4 modelslast run Aug 18, 2026
Quality
77.0%
Claude Opus 5 · high
Value
$0.10
GPT-5.6 Luna · xhigh
Speed
1.6m
GPT-5.6 Luna · xhigh
HLE
Humanity's Last Exam as a search benchmark: expert questions answered with live search.
2 modelslast run Aug 17, 2026
Quality
77.4%
Claude Opus 5 · high
Value
$0.16
Claude Opus 5 · high
Speed
48s
Claude Opus 5 · high
WideSearch
Fill an entire table; answer-item accuracy scores partial matches.
4 modelslast run Aug 18, 2026
Quality
84.0%
GPT-5.6 Sol · high
Value
$0.063
GPT-5.6 Luna · xhigh
Speed
1.9m
GPT-5.6 Sol · high
Benchmark results are a dated demonstration dataset. Explore the source benchmark index for live results.