Intel · knowledge base · updated daily
The current state of AI, in one place.
Not a news feed — a live picture. Models and what they cost, who supports what, which tools fit your industry, and what's rising. Updated every day; recent changes called out.
Models — cost and capabilityread free · compare & export need an account
| Change | Model | Context | $ in /M | $ out /M | Coding | Reasoning | Updated |
|---|---|---|---|---|---|---|---|
| ↑ new | Claude Fable 5 | 500K | 5.00 | 25.00 | 86.3 | 94.0 | 07-05 |
| ▲ changed | Claude Opus 4.8 | 500K | 4.00 | 20.00 | 82.1 | 91.4 | 07-04 |
| — stable | Claude Sonnet 5 | 400K | 1.50 | 7.50 | 78.9 | 87.2 | 07-02 |
| — stable | GPT-5.2 | 256K | 3.00 | 15.00 | 79.4 | 88.0 | 06-18 |
| — stable | Gemini 3 Pro | 1M | 2.50 | 12.00 | 76.8 | 86.1 | 06-11 |
every number carries its source on click · unverified items are marked · ⊘ compare selected · CSV export · row alerts — free account
Who supports what● yes · ◐ partial · ○ no
Tool use
Vision
Long docs
On-device
Anthropic
●
●
●
○
OpenAI
●
●
◐
○
Google
●
●
●
◐
Meta
◐
◐
○
●
Tools by industrywhat fits your work
Trades & field service
Scheduling + intake automation is the mature use; voice agents rising fast.
as of 07-05 · source on click
Law & professional services
Document drill and intake prep are dependable; billing automation still early.
as of 07-05 · source on click
Education
Tutoring agents with real work-checking are the frontier — few do it well.
as of 07-05 · source on click
What's risingagents · skills · platforms
1Long-running coding agentsmulti-hour autonomous builds↑ rising
2Sub-agent orchestrationfan-out review patterns↑ rising
3MCP tool serversinterop standard settlingnew
1Prompt cachingcost lever most teams miss— steady
2Computer useagents driving real screens↑ rising
3Voice-first agentsfield-work entry pointnew