Custom providers on Manifest
24.4B tokens
/ 30 days
39 models routed through Manifest over the last 30 days.
Bring any OpenAI-compatible endpoint — vLLM, Ollama, LiteLLM, LM Studio, or a private gateway — and Manifest routes it like any other provider, with a spending cap per agent and automatic fallback. Every custom endpoint connected to Manifest is grouped on this page as one provider.
Custom provider tokens processed on Manifest
Models in use
# Model |
Auth type | Tokens / last 30 days |
|---|---|---|
1.MiniMax M3 | API Key | 9.3 B |
2.Other Custom Models | API Key | 4.5 B |
3.Claude Opus 4 8 | API Key | 2.8 B |
4.Claude Sonnet 4 6 | API Key | 1.4 B |
5.Claude Sonnet 4 6 | API Key | 1.2 B |
6.Mimo V2.5 Pro Free | API Key | 761 M |
7.Mimo V2.5 Pro | API Key | 633 M |
8.Agnes 2.0 Flash | API Key | 618 M |
9.Deepseek V4 Flash Free | API Key | 539 M |
10.Kimi K2.6 | API Key | 348 M |
11.Mimo V2.5 Free | API Key | 346 M |
12.Claude Opus 4 6 | API Key | 250 M |
13.Nemotron 3 Super 120b A12b | API Key | 249 M |
14.Deepseek V4 Flash | API Key | 247 M |
15.Minimax M3 | API Key | 238 M |
16.Gpt 5.5 | API Key | 199 M |
17.Minimax M2.7 | API Key | 134 M |
18.Openrouter Free | API Key | 95 M |
19.Claude Opus 4 7 | API Key | 79 M |
20.Kimi K2.6 | API Key | 75 M |
21.Claude Haiku 4 5 20251001 | API Key | 70 M |
22.Deepseek V4 Pro | API Key | 67 M |
23.Big Pickle | API Key | 53 M |
24.Glm 5.2 | API Key | 49 M |
25.Glm 5.2 Free | API Key | 42 M |
26.Kimi K2.6 | API Key | 40 M |
27.Glm 5.1 | API Key | 33 M |
28.Claude Sonnet 4.5 | API Key | 21 M |
29.Gpt 5.4 | API Key | 18 M |
30.Gpt 5.4 Mini | API Key | 15 M |
31.Minimax M3 Free | API Key | 11 M |
32.Nemotron 3 Nano Omni 30b A3b Reasoning | API Key | 9 M |
33.Gpt Oss 120b | API Key | 8 M |
34.Gpt 5.3 Codex | API Key | 5 M |
35.Claude Fable 5 | API Key | 3 M |
36.Zai Glm 4.7 | API Key | 3 M |
37.Gpt 5.4 | API Key | 2 M |
38.Deepseek V3.2 | API Key | 1 M |
39.Gpt Oss 120b | API Key | 127 K |
Model names are aggregated across every connected custom endpoint. Models used by only a handful of workspaces are grouped under “Other Custom Models” to keep individual configurations private.