benchr
Featured today
GPT-5 versus Claude Opus 4.7: seven workload decision factors
An editorial guide to code review, landing-page copy, citation checks, recipes, and email drafting, grounded in cited public evidence and a reproducible local test plan.
The decision platform for AI models
Source-backed model intelligence, your own evaluation cases, production cost scenarios, and migration planning — connected in one workspace.
AI model intelligence workspace
Explore. Test. Decide. Ship.
Go from scattered model claims to a decision you can defend. Search verified model and API data, build a shortlist, run your own evaluation set, estimate the bill, and plan migrations before a model retires.
Explore
Model & API catalog
Filter the live index, inspect evidence, shortlist candidates, and save a decision workspace.
Open catalog → LabsBring-your-output evals
Test 2–5 models with your prompts, blind-score outputs, and export a reproducible report.
Start an eval → EconomicsWorkload cost lab
Model real traffic with cached input, batch pricing, and output-heavy agent scenarios.
Price a workload → OperationsMigration assistant
Find a retired API ID, compare its official replacement, and generate a migration checklist.
Plan a migration →27verified models
25logged changes
41retirement records
4 labelsevidence categories
Live record
Latest model changes
LoadingReading the verified change record…
Latest
-
01
GPT-5.6: Sol, Terra, Luna
Sol, Terra, and Luna reached general availability on July 9. Sol now lists a 1.05M context window and $5/$30 standard pricing; the article preserves the preview-to-GA correction trail.
-
02
Claude Fable 5
The $10/$50 Mythos-class model is available worldwide again after its June 12–30 suspension; access was restored July 1. Safety-classified requests refuse explicitly, with another-model retry only when an app configures it.
-
03
Claude Opus 4.8
Same price as 4.7, a small leaderboard bump, and an honesty gain that catches its own bugs.
-
04
GPT-5.5
OpenAI's gains land in agentic coding and computer use, at roughly double GPT-5's API price.
-
05
Opus 4.8 vs GPT-5.5
Both list $5 per million input tokens. Compare their published specifications, then validate the fit on your own stack.
-
06
Claude Mythos
Anthropic's restricted frontier model, gated under Project Glasswing. What it is and why it's locked away.
Reviews
-
07
Claude Opus 4.7
Strong on architecture decisions and long-document analysis. Vision is its weaker side.
-
08
GPT-5
A candidate for visual and structured-output work based on OpenAI's published positioning; validate it on your own prompts.
-
09
Gemini 3 Pro
A retired multimodal model that Google deprecated in March 2026 in favor of Gemini 3.1 Pro.
Comparisons
-
10
Coding assistants
Cursor, GitHub Copilot, Windsurf, and Cody on the same Markdown-exporter task.
-
11
Voice models
ElevenLabs, OpenAI Whisper, and Cartesia Sonic on latency, accuracy, and naturalness.
-
12
Context windows
Advertised context capacity versus reproducible retrieval checks you can run on your own documents.
Analysis
-
13
Price per use case
The cheapest model for chat, coding, RAG, agents, classification, and summarization.
-
14
The open-weight tier
Where Llama 4, Mistral Large 2, DeepSeek-V3.1, and Qwen 3 stand against the closed labs.
-
15
Why benchmarks stopped telling you
MMLU is saturated above 90%. The benchmarks worth tracking now.
Guides
-
16
Frontier models
Claude Opus 4.7, GPT-5, and Gemini 3.1 Pro Preview: sourced specifications, editorial tradeoffs, and a local evaluation plan.
-
17
Open-weight models
Llama 4, Mistral, DeepSeek, Qwen, the small-model tier, and what it takes to self-host.
-
18
AI costs
What AI costs by model and workload, and where teams overspend.
Compare models directly
An interactive comparison covering pricing, benchmarks, context windows, and capability ratings for all 27 models in the shared index, across frontier, mid, and open-weight tiers.
All 27 models ranked by benchr Rating →
benchr is an evidence-led reference, not a generic listicle. Figures are identified as official provider data, third-party benchmark results, or benchr editorial estimates so readers can judge the evidence behind each claim.
- Traceable evidence. Official prices and specifications link to provider documentation; benchmark results link to their maintainers; estimates and analysis are labeled as such. Verification dates and source fields are published in our open data file.
- A record no one else keeps. An append-only price history and a permanent deprecation register; retired models are never deleted — that memory is the point.
- Editorially independent and commercially transparent. Some marked links to AI/ML API, Lovable, and RunPod can earn benchr a commission without changing your price. Partners do not buy rankings or approve coverage. Google AdSense may be added only after approval. Read the affiliate disclosure, editorial standards, and corrections policy.