Skip to content

Commit 11221db

Browse files
ErikBjareErikBjare
committed
refactor: moved openai model metadata into seperate file, added make update-models to use gptme to update it
1 parent f2b8be1 commit 11221db

File tree

4 files changed

+153
-65
lines changed

4 files changed

+153
-65
lines changed

‎Makefile‎

Lines changed: 4 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -41,6 +41,10 @@ format:
4141
poetry run pyupgrade --py310-plus --exit-zero-even-if-changed ${SRCFILES}
4242
poetry run black ${SRCDIRS}
4343

44+
update-models:
45+
wayback_url=$$(curl "https://archive.org/wayback/available?url=openai.com/api/pricing/" | jq -r '.archived_snapshots.closest.url') && \
46+
gptme 'update the model metadata from this page' gptme/models.py gptme/llm_openai_models.py "$${wayback_url}" --non-interactive
47+
4448
precommit: format lint typecheck test
4549

4650
docs/.clean: docs/conf.py

‎gptme/llm_openai.py‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -4,6 +4,7 @@
44

55
from .constants import TEMPERATURE, TOP_P
66
from .message import Message, msgs2dicts
7+
from .models import ModelMeta
78

89
if TYPE_CHECKING:
910
from openai import OpenAI

‎gptme/llm_openai_models.py‎

Lines changed: 145 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,145 @@
1+
from typing import TypedDict
2+
from typing_extensions import NotRequired
3+
4+
class _ModelDictMeta(TypedDict):
5+
context: int
6+
max_output: NotRequired[int]
7+
price_input: NotRequired[float]
8+
price_output: NotRequired[float]
9+
10+
OPENAI_MODELS: dict[str, _ModelDictMeta] = {
11+
# GPT-4o
12+
"gpt-4o": {
13+
"context": 128_000,
14+
"price_input": 5,
15+
"price_output": 15,
16+
},
17+
"gpt-4o-2024-08-06": {
18+
"context": 128_000,
19+
"price_input": 2.5,
20+
"price_output": 10,
21+
},
22+
"gpt-4o-2024-05-13": {
23+
"context": 128_000,
24+
"price_input": 5,
25+
"price_output": 15,
26+
},
27+
# GPT-4o mini
28+
"gpt-4o-mini": {
29+
"context": 128_000,
30+
"price_input": 0.15,
31+
"price_output": 0.6,
32+
},
33+
"gpt-4o-mini-2024-07-18": {
34+
"context": 128_000,
35+
"price_input": 0.15,
36+
"price_output": 0.6,
37+
},
38+
# OpenAI o1-preview
39+
"o1-preview": {
40+
"context": 128_000,
41+
"price_input": 15,
42+
"price_output": 60,
43+
},
44+
"o1-preview-2024-09-12": {
45+
"context": 128_000,
46+
"price_input": 15,
47+
"price_output": 60,
48+
},
49+
# OpenAI o1-mini
50+
"o1-mini": {
51+
"context": 128_000,
52+
"price_input": 3,
53+
"price_output": 12,
54+
},
55+
"o1-mini-2024-09-12": {
56+
"context": 128_000,
57+
"price_input": 3,
58+
"price_output": 12,
59+
},
60+
# GPT-4 Turbo
61+
"gpt-4-turbo": {
62+
"context": 128_000,
63+
"price_input": 10,
64+
"price_output": 30,
65+
},
66+
"gpt-4-turbo-2024-04-09": {
67+
"context": 128_000,
68+
"price_input": 10,
69+
"price_output": 30,
70+
},
71+
"gpt-4-0125-preview": {
72+
"context": 128_000,
73+
"price_input": 10,
74+
"price_output": 30,
75+
},
76+
"gpt-4-1106-preview": {
77+
"context": 128_000,
78+
"price_input": 10,
79+
"price_output": 30,
80+
},
81+
"gpt-4-vision-preview": {
82+
"context": 128_000,
83+
"price_input": 10,
84+
"price_output": 30,
85+
},
86+
# GPT-4
87+
"gpt-4": {
88+
"context": 8192,
89+
"price_input": 30,
90+
"price_output": 60,
91+
},
92+
"gpt-4-32k": {
93+
"context": 32768,
94+
"price_input": 60,
95+
"price_output": 120,
96+
},
97+
# GPT-3.5 Turbo
98+
"gpt-3.5-turbo-0125": {
99+
"context": 16385,
100+
"price_input": 0.5,
101+
"price_output": 1.5,
102+
},
103+
"gpt-3.5-turbo": {
104+
"context": 16385,
105+
"price_input": 0.5,
106+
"price_output": 1.5,
107+
},
108+
"gpt-3.5-turbo-instruct": {
109+
"context": 4096,
110+
"price_input": 1.5,
111+
"price_output": 2,
112+
},
113+
# Deprecated models (kept for reference)
114+
"gpt-3.5-turbo-1106": {
115+
"context": 16385,
116+
"price_input": 1,
117+
"price_output": 2,
118+
},
119+
"gpt-3.5-turbo-0613": {
120+
"context": 4096,
121+
"price_input": 1.5,
122+
"price_output": 2,
123+
},
124+
"gpt-3.5-turbo-16k-0613": {
125+
"context": 16385,
126+
"price_input": 3,
127+
"price_output": 4,
128+
},
129+
"gpt-3.5-turbo-0301": {
130+
"context": 4096,
131+
"price_input": 1.5,
132+
"price_output": 2,
133+
},
134+
# Other models
135+
"davinci-002": {
136+
"context": 4096, # Assuming default context size
137+
"price_input": 2,
138+
"price_output": 2,
139+
},
140+
"babbage-002": {
141+
"context": 4096, # Assuming default context size
142+
"price_input": 0.4,
143+
"price_output": 0.4,
144+
},
145+
}

‎gptme/models.py‎

Lines changed: 3 additions & 65 deletions
Original file line numberDiff line numberDiff line change
@@ -4,6 +4,8 @@
44

55
from typing_extensions import NotRequired
66

7+
from .llm_openai_models import OPENAI_MODELS
8+
79
logger = logging.getLogger(__name__)
810

911

@@ -38,71 +40,7 @@ class _ModelDictMeta(TypedDict):
3840
# known models metadata
3941
# TODO: can we get this from the API?
4042
MODELS: dict[str, dict[str, _ModelDictMeta]] = {
41-
"openai": {
42-
# gpt-3.5
43-
"gpt-3.5-turbo": {
44-
"context": 4097,
45-
"price_input": 1,
46-
"price_output": 2,
47-
},
48-
"gpt-3.5-turbo-16k": {
49-
"context": 16385,
50-
},
51-
"gpt-3.5-turbo-1106": {
52-
"context": 16385,
53-
},
54-
# gpt-4
55-
"gpt-4": {
56-
"context": 8193,
57-
"price_input": 30,
58-
"price_output": 60,
59-
},
60-
"gpt-4-32k": {
61-
"context": 32769,
62-
"price_input": 60,
63-
"price_output": 120,
64-
},
65-
# gpt-4-turbo
66-
# https://openai.com/blog/new-models-and-developer-products-announced-at-devday
67-
"gpt-4-1106-preview": {
68-
"context": 128_000,
69-
},
70-
"gpt-4-vision-preview": {
71-
"context": 128_000,
72-
},
73-
"gpt-4-turbo": {
74-
"context": 128_000,
75-
"price_input": 10,
76-
"price_output": 30,
77-
},
78-
"gpt-4o": {
79-
"context": 128_000,
80-
"price_input": 5,
81-
"price_output": 15,
82-
},
83-
"gpt-4o-2024-08-06": {
84-
"context": 128_000,
85-
"price_input": 2.5,
86-
"price_output": 10,
87-
},
88-
"gpt-4o-mini": {
89-
"context": 128_000,
90-
"price_input": 0.15,
91-
"price_output": 0.6,
92-
},
93-
"o1-mini": {
94-
"context": 128_000,
95-
"max_output": 65536,
96-
"price_input": 3,
97-
"price_output": 12,
98-
},
99-
"o1-preview": {
100-
"context": 128_000,
101-
"max_output": 32768,
102-
"price_input": 15,
103-
"price_output": 60,
104-
},
105-
},
43+
"openai": OPENAI_MODELS,
10644
"anthropic": {
10745
"claude-3-opus-20240229": {
10846
"context": 200_000,

0 commit comments

Comments
 (0)