llmr 0.2.0

Reach language models across providers, with capabilities you can read before you ask and usage you can trust.
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110

# The OpenAI models this release knows about.
#
# A starting point, not a source of truth. Every row was read off a vendor model page on the
# date it carries, and vendors add and retire models on their own schedule. Supply your own
# Registry when you need one that is current: `Registry::stale` and `Registry::unlisted`
# compare this table against what the API says it serves.
#
# This table is for OpenAI's own endpoint. The `openai` provider in this crate speaks a shape
# that a dozen vendors answer, and their models are not these. Build a Registry of your own
# for a Groq or an Ollama, and give it the reach that is true of where it runs.
#
# `prompt_caching` and `thinking` on the mini and nano rows are not in the feature list on
# their model pages. They are here because the pricing page publishes a cached input rate for
# both, and the model pages carry "reasoning token support" in their details. That is
# evidence read off a published page rather than an assumption, and it is written down here
# so the next person can check the same two places.

provider = "openai"
reach    = "FirstPartyApi"

[[model]]
id                = "gpt-5.6-sol"
context_window    = 1050000
max_output        = 128000
tools             = true
structured_output = true
prompt_caching    = true
thinking          = true
images            = true
streaming         = true
source            = "vendor model page"
verified_at       = "2026-08-31"

[[model]]
id                = "gpt-5.6-terra"
context_window    = 1050000
max_output        = 128000
tools             = true
structured_output = true
prompt_caching    = true
thinking          = true
images            = true
streaming         = true
source            = "vendor model page"
verified_at       = "2026-08-31"

[[model]]
id                = "gpt-5.6-luna"
context_window    = 1050000
max_output        = 128000
tools             = true
structured_output = true
prompt_caching    = true
thinking          = true
images            = true
streaming         = true
source            = "vendor model page"
verified_at       = "2026-08-31"

[[model]]
id                = "gpt-5.1"
context_window    = 400000
max_output        = 128000
tools             = true
structured_output = true
prompt_caching    = true
thinking          = true
images            = true
streaming         = true
source            = "vendor model page"
verified_at       = "2026-08-31"

[[model]]
id                = "gpt-5"
context_window    = 400000
max_output        = 128000
tools             = true
structured_output = true
prompt_caching    = true
thinking          = true
images            = true
streaming         = true
source            = "vendor model page"
verified_at       = "2026-08-31"

[[model]]
id                = "gpt-5-mini"
context_window    = 400000
max_output        = 128000
tools             = true
structured_output = true
prompt_caching    = true
thinking          = true
images            = true
streaming         = true
source            = "vendor model page and pricing page"
verified_at       = "2026-08-31"

[[model]]
id                = "gpt-5-nano"
context_window    = 400000
max_output        = 128000
tools             = true
structured_output = true
prompt_caching    = true
thinking          = true
images            = true
streaming         = true
source            = "vendor model page and pricing page"
verified_at       = "2026-08-31"