01 / MODELS
Choose a model
Find current model IDs, prices, context limits and provider access routes.
Browse model pages →Independent LLM API intelligence
Choose the right LLM API. With current, source-checked data.
Compare models and providers by price, context, capabilities and source-dated evidence—then go directly to the provider that fits your workload.
50model & API pages
61decision comparisons
25provider profiles
173indexed research URLs
Current site inventory snapshot · verified 2026-08-23 · coverage count, not adoption or customer usage
Decision workspace
Current facts you can act on.
A sample of provider-published standard API prices. Prices are USD per 1M text tokens and exclude batch, cache, fast-mode, regional and tool fees unless stated.
| Model / provider | Useful for | Input / output | Context | Evidence | Source |
|---|---|---|---|---|---|
GPT-5.6 Terra OpenAI |
Balanced intelligence and cost | $2 / $12 | 1.05M | Checked 2026-08-23 | Official source ↗ |
Claude Sonnet 5 Anthropic |
Agentic and coding workloads | $2 / $10 through 2026-08-31; then $3 / $15 |
Not stated on pricing source | Checked 2026-08-23 | Official source ↗ |
Gemini 3.1 Pro Preview Google |
Multimodal, agentic, long-context work | $2 / $12 ≤200k prompt; $4 / $18 above |
1M input | Checked 2026-08-23 | Official source ↗ |
Source hierarchy: provider docs → provider announcements → reproducible measurementOpen the full pricing comparison →
Start from the decision
Three ways into the data.
No signup gate. Start with the question you need to answer and inspect the evidence before you integrate.
02 / COMPARE
Compare providers
Evaluate cost, capabilities, limits and operational trade-offs side by side.
Open comparisons →03 / VERIFY
Verify a price
Check the latest published rate, its conditions and the date we reviewed it.
Verify API pricing →How we verify
Clear evidence states, not false precision.
The directory is useful only if a buyer can distinguish a provider fact, a VerticalAPI measurement and a missing field.
01
First-party sources
Provider pricing, model and policy pages come first. Secondary reporting does not override a primary source.
02
Dated evidence
Facts that can drift carry a checked date and conditions such as context tier, region or preview status.
03
“Not published”
If a provider does not publish a figure, the field stays incomplete. Missing coverage is never converted to zero.
04
Corrections
Providers and readers can submit a source. Accepted corrections update the evidence, never an advertiser’s preferred wording.
Research hubs
Follow the questions buyers already ask.
Editorial routes connect model pages, provider profiles, benchmarks and specific workload decisions.
Benchmark protocol
Latency, throughput, quality and list cost across LLM providers.
The protocol is public now; results stay pending until a complete run is reproducible and its limitations are documented.
For API providers
Make your listing accurate. Not flattering.
Submit a new API, correct a published fact or provide a first-party source. Basic listings and factual corrections are free. Sponsored visibility, if introduced, is visibly labelled and never changes editorial rankings.
BETA UTILITY
Need an OpenAI-compatible BYOK gateway?
The gateway remains available as a secondary developer utility while the intelligence directory is the primary VerticalAPI product. Availability and access are handled directly.