Commit Graph
29 Commits
Author SHA1 Message Date
inference-bot 6b41382bef models: Regolo's 2026-11-01 prices — apertus-70b $1.16/$3.94, qwen3.5-122b $0.70/$3.94 2026-11-01 00:44:07 +00:00
inference-bot 2fdda5a201 privacy: state what the API path actually stores (and add it where members read)
The policy described chat history and enclave models carefully but never addressed
the API path, and §2 read as though all content is retained — which is wrong for
members using their own keys.

Verified before writing, rather than restating policy language: a request with a
distinctive marker was sent through the gateway and searched for afterwards. It
appears nowhere in the gateway's logs, `litellm_settings` enables no content
logging, and the request records carry usage fields only (model, tokens, cost, key).

- privacy-policy.md §2: content retention now depends on the surface, with a pointer.
- privacy-policy.md §4: new subsection on API requests, including the two limits that
  stop it being read as more than it says — we can still link a request to a key and
  therefore a member, and "not stored by us" says nothing about the provider.
- api-access.md: a "what we store" section, since that is the page an API user reads
  before worrying about it.
- models.md: the privacy-limitations section now distinguishes chat from API.
2026-10-29 18:21:01 +00:00
inference-bot e0732af1fd models: apertus-70b is 262K context, not 65K — measured, not copied
The provider's published catalog says this model takes 30K input and has no
vision support. Both of those are wrong. Sent through our own gateway:

- an image returns the right answer ("Red" for a solid red PNG)
- prompts of 32K, 70K, 130K and 262K input tokens are all accepted, with
  Regolo's own usage accounting reporting prompt_tokens=262021

Our own description of the v1.5 generation — image understanding and a 262K
context window — matches what the model demonstrably does, which is why the
figure goes to 262K. The old 65K undersold the model by roughly 4x.

Regolo's meta-model for qwen3.5-122b says 120K; it processed 262,023 tokens and
still followed the instruction at the end, so that row's 262K is confirmed as
written and is left alone.

Adds a line under the table recording that context figures here are measured
through the gateway rather than copied from provider catalogs, because that
difference is exactly what caused a member to report this as an error.
2026-10-29 17:11:17 +00:00
inference-bot 6cf50e6a5d docs: link the energy and carbon page from the model list 2026-10-29 04:16:48 +00:00
inference-bot e7c86b16ba models: fix incorrect 'biggest context' claim for qwen3.5-122b (issue #4 — GLM/DeepSeek have 1M) 2026-10-16 16:32:08 -06:00
inference-bot d3c3c8f8f7 models: normalize Green column to ✅ format; explain privacy column terms (architectural/policy/ZDR) 2026-10-16 11:05:11 -06:00
inference-bot 884340427f models: dedupe green emoji legend; fix image-gen instructions (Integrations button > Image) 2026-10-16 10:59:05 -06:00
inference-bot 9699ff08cc models: remove PublicAI provider; Regolo becomes the Apertus provider; emoji = feature tier; add Qwen-Image image-generation section 2026-10-16 10:50:46 -06:00
inference-bot 0d4a62a2da models: add Regolo provider (Apertus 70B, Qwen 3.5 122B) + image generation via regolo/qwen-image 2026-10-16 10:29:48 -06:00
inference-bot a5e33b73dd Swap tinfoil/glm-5-3-flash (deprecated 10-09) for tinfoil/glm-5.3 flagship 2026-10-07 22:08:48 -06:00
Sebastian Dröge 3c346e420f models: Correctly document greenpt/glm-5.3 cache read pricing 2026-10-05 09:26:01 +03:00
inference-bot c022a0b52e models: add cached-input pricing column (cache-read rates per model) 2026-10-03 09:34:32 -06:00
inference-admin 85f5723357 added note about the European Open Source AI Index 2026-10-03 04:42:02 +00:00
inference-bot a647284170 models: remove phantom tinfoil/glm-5-3 and kimi-k3 (not served); reconcile to gateway catalog 2026-10-02 22:36:28 -06:00
inference-bot 888ac8c299 models: value-tier emoji (🔒/🌱/📖) matching the chat picker 2026-10-02 22:31:51 -06:00
inference-bot dcdde204b0 models: greenpt/deepseek-v4.1-flash as default; Tinfoil deepseek as TEE twin 2026-10-02 22:15:47 -06:00
inference-bot 0a79c80483 docs: remove green-l/green-r from model list and API examples 2026-10-02 22:04:51 -06:00
inference-bot 5f0c7b4513 Models page: vision column (verified), context windows, provider education, how-to-pick guide, same-model-two-providers note 2026-09-29 12:16:31 -06:00
inference-bot 34ce3b6753 Document audio models (STT/TTS) with API examples 2026-09-27 11:25:50 -06:00
inference-bot 40925b82b5 Add greenpt/glm-5.3 to model list 2026-09-26 16:52:24 -06:00
inference-bot 4710df17ef Add PublicAI Apertus models + sovereignty/open-source notes 2026-09-25 15:56:12 -06:00
inference-bot 6a005947a9 Merge naming and provider comparison into one Providers table 2026-09-24 11:49:38 -06:00
inference-bot f3007e4c7b Add provider comparison chart with privacy and green-energy gradations 2026-09-24 10:56:10 -06:00
inference-bot 27d41a840e Add greenpt/glm-5.3-flash to model list 2026-09-24 10:43:18 -06:00
inference-bot faeb958232 Update model names to tinfoil/* prefix 2026-09-24 08:16:19 -06:00
inference-bot a92251d80e Note models.md pricing mirrors LiteLLM config (kept in sync) 2026-09-23 21:08:30 -06:00
inference-bot 9fd074aaa7 Add per-model pricing (in/out per 1M tokens) to the model list 2026-09-23 21:07:04 -06:00
inference-bot 811da1617a Rename privacy caveat heading to 'Limitations on privacy' 2026-09-23 21:06:09 -06:00
inference-bot c6b659aeef Extract model catalog into a single models.md; point chat + API docs at it 2026-09-23 20:47:38 -06:00