Skip to content

Commit f3b10b7

Browse files
haraldschillyharaldschilly
committed
util/llm: update pricing and names
1 parent 5d9a838 commit f3b10b7

File tree

2 files changed

+49
-49
lines changed

2 files changed

+49
-49
lines changed

‎src/packages/util/db-schema/llm-utils.ts

Lines changed: 33 additions & 33 deletions
Original file line numberDiff line numberDiff line change
@@ -700,32 +700,32 @@ export const LLM_USERNAMES: LLM2String = {
700700
"gpt-3.5-turbo": "GPT-3.5",
701701
"gpt-3.5-turbo-16k": "GPT-3.5-16k",
702702
"gpt-4-turbo-preview": "GPT-4 Turbo 128k",
703-
"gpt-4-turbo-preview-8k": "GPT-4 Turbo 8k",
703+
"gpt-4-turbo-preview-8k": "GPT-4 Turbo",
704704
"gpt-4-turbo": "GPT-4 Turbo 128k",
705-
"gpt-4-turbo-8k": "GPT-4 Turbo 8k",
706-
"gpt-4o": "GPT-4 Omni 128k",
707-
"gpt-4o-8k": "GPT-4 Omni 8k",
705+
"gpt-4-turbo-8k": "GPT-4 Turbo",
706+
"gpt-4o": "GPT-4o 128k",
707+
"gpt-4o-8k": "GPT-4o",
708708
"gpt-4o-mini": "GPT-4o Mini 128k",
709-
"gpt-4o-mini-8k": "GPT-4o Mini 8k",
709+
"gpt-4o-mini-8k": "GPT-4o Mini",
710710
"text-embedding-ada-002": "Text Embedding Ada 002", // TODO: this is for embeddings, should be moved to a different place
711711
"text-bison-001": "PaLM 2",
712712
"chat-bison-001": "PaLM 2",
713713
"gemini-pro": "Gemini 1.0 Pro",
714714
"gemini-1.0-ultra": "Gemini 1.0 Ultra",
715715
"gemini-1.5-pro": "Gemini 1.5 Pro 1m",
716-
"gemini-1.5-pro-8k": "Gemini 1.5 Pro 8k",
717-
"gemini-1.5-flash-8k": "Gemini 1.5 Flash 8k",
716+
"gemini-1.5-pro-8k": "Gemini 1.5 Pro",
717+
"gemini-1.5-flash-8k": "Gemini 1.5 Flash",
718718
"mistral-small-latest": "Mistral AI Small",
719719
"mistral-medium-latest": "Mistral AI Medium",
720720
"mistral-large-latest": "Mistral AI Large",
721-
"claude-3-haiku": "Claude 3 Haiku",
722-
"claude-3-haiku-8k": "Claude 3 Haiku 8k",
723-
"claude-3-sonnet": "Claude 3 Sonnet",
724-
"claude-3-sonnet-4k": "Claude 3 Sonnet 4k",
725-
"claude-3-5-sonnet": "Claude 3.5 Sonnet",
726-
"claude-3-5-sonnet-4k": "Claude 3.5 Sonnet 4k",
721+
"claude-3-haiku": "Claude 3 Haiku 200k",
722+
"claude-3-haiku-8k": "Claude 3 Haiku",
723+
"claude-3-sonnet": "Claude 3 Sonnet 200k",
724+
"claude-3-sonnet-4k": "Claude 3 Sonnet",
725+
"claude-3-5-sonnet": "Claude 3.5 Sonnet 200k",
726+
"claude-3-5-sonnet-4k": "Claude 3.5 Sonnet",
727727
"claude-3-opus": "Claude 3 Opus 200k",
728-
"claude-3-opus-8k": "Claude 3 Opus 8k",
728+
"claude-3-opus-8k": "Claude 3 Opus",
729729
} as const;
730730

731731
// similar to the above, we map to short user-visible description texts
@@ -736,17 +736,17 @@ export const LLM_DESCR: LLM2String = {
736736
chatgpt4:
737737
"Can follow complex instructions and solve difficult problems. (OpenAI, 8k token context)",
738738
"gpt-4":
739-
"Most powerful OpenAI model. Can follow complex instructions and solve difficult problems. (OpenAI, 8k token context)",
739+
"Powerful OpenAI model. Can follow complex instructions and solve difficult problems. (OpenAI, 8k token context)",
740740
"gpt-4-32k": "",
741741
"gpt-3.5-turbo": "Fast, great for everyday tasks. (OpenAI, 4k token context)",
742742
"gpt-3.5-turbo-16k": `Same as ${LLM_USERNAMES["gpt-3.5-turbo"]} but with larger 16k token context`,
743743
"gpt-4-turbo-preview-8k":
744744
"More powerful, fresher knowledge, and lower price than GPT-4. (OpenAI, 8k token context)",
745745
"gpt-4-turbo-preview":
746-
"Like GPT-4 Turbo 8k, but with up to 128k token context",
746+
"Like GPT-4 Turbo, but with up to 128k token context",
747747
"gpt-4-turbo-8k":
748748
"Faster, fresher knowledge, and lower price than GPT-4. (OpenAI, 8k token context)",
749-
"gpt-4-turbo": "Like GPT-4 Turbo 8k, but with up to 128k token context",
749+
"gpt-4-turbo": "Like GPT-4 Turbo, but with up to 128k token context",
750750
"gpt-4o-8k":
751751
"Most powerful, fastest, and cheapest (OpenAI, 8k token context)",
752752
"gpt-4o": "Most powerful fastest, and cheapest (OpenAI, 128k token context)",
@@ -873,14 +873,14 @@ export const LLM_COST: { [name in LanguageModelCore]: Cost } = {
873873
free: false,
874874
},
875875
"gpt-3.5-turbo": {
876-
prompt_tokens: usd1Mtokens(1.5),
877-
completion_tokens: usd1Mtokens(2),
876+
prompt_tokens: usd1Mtokens(3),
877+
completion_tokens: usd1Mtokens(6),
878878
max_tokens: 4096,
879879
free: true,
880880
},
881881
"gpt-3.5-turbo-16k": {
882-
prompt_tokens: usd1Mtokens(0.5),
883-
completion_tokens: usd1Mtokens(1.5),
882+
prompt_tokens: usd1Mtokens(3),
883+
completion_tokens: usd1Mtokens(6),
884884
max_tokens: 16384,
885885
free: false,
886886
},
@@ -910,14 +910,14 @@ export const LLM_COST: { [name in LanguageModelCore]: Cost } = {
910910
free: false,
911911
},
912912
"gpt-4o-8k": {
913-
prompt_tokens: usd1Mtokens(5),
914-
completion_tokens: usd1Mtokens(15),
913+
prompt_tokens: usd1Mtokens(2.5),
914+
completion_tokens: usd1Mtokens(10),
915915
max_tokens: 8192, // like gpt-4-turbo-8k
916916
free: false,
917917
},
918918
"gpt-4o": {
919-
prompt_tokens: usd1Mtokens(5),
920-
completion_tokens: usd1Mtokens(15),
919+
prompt_tokens: usd1Mtokens(2.5),
920+
completion_tokens: usd1Mtokens(10),
921921
max_tokens: 128000, // This is a lot: blows up the "max cost" calculation → requires raising the minimum balance and quota limit
922922
free: false,
923923
},
@@ -966,15 +966,15 @@ export const LLM_COST: { [name in LanguageModelCore]: Cost } = {
966966
free: true,
967967
},
968968
"gemini-1.5-flash-8k": {
969-
prompt_tokens: usd1Mtokens(0.35),
970-
completion_tokens: usd1Mtokens(1.05),
969+
prompt_tokens: usd1Mtokens(0.075),
970+
completion_tokens: usd1Mtokens(0.3),
971971
max_tokens: 8_000,
972972
free: true,
973973
},
974974
// https://mistral.ai/technology/
975975
"mistral-small-latest": {
976-
prompt_tokens: usd1Mtokens(1),
977-
completion_tokens: usd1Mtokens(3),
976+
prompt_tokens: usd1Mtokens(0.2),
977+
completion_tokens: usd1Mtokens(0.6),
978978
max_tokens: 4096, // TODO don't know the real value, see getMaxTokens
979979
free: true,
980980
},
@@ -985,8 +985,8 @@ export const LLM_COST: { [name in LanguageModelCore]: Cost } = {
985985
free: true,
986986
},
987987
"mistral-large-latest": {
988-
prompt_tokens: usd1Mtokens(4),
989-
completion_tokens: usd1Mtokens(12),
988+
prompt_tokens: usd1Mtokens(2),
989+
completion_tokens: usd1Mtokens(6),
990990
max_tokens: 4096, // TODO don't know the real value, see getMaxTokens
991991
free: false,
992992
},
@@ -1006,13 +1006,13 @@ export const LLM_COST: { [name in LanguageModelCore]: Cost } = {
10061006
"claude-3-5-sonnet": {
10071007
prompt_tokens: usd1Mtokens(3),
10081008
completion_tokens: usd1Mtokens(15),
1009-
max_tokens: 4_000, // limited to 4k tokens, offered for free
1009+
max_tokens: 200_000,
10101010
free: false,
10111011
},
10121012
"claude-3-5-sonnet-4k": {
10131013
prompt_tokens: usd1Mtokens(3),
10141014
completion_tokens: usd1Mtokens(15),
1015-
max_tokens: 4_000, // limited to 4k tokens, offered for free
1015+
max_tokens: 4_000, // limited to 4k tokens
10161016
free: false,
10171017
},
10181018
"claude-3-sonnet-4k": {

‎src/packages/util/db-schema/purchase-quotas.ts

Lines changed: 16 additions & 16 deletions
Original file line numberDiff line numberDiff line change
@@ -21,36 +21,36 @@ interface Spec {
2121

2222
export type QuotaSpec = Record<Service, Spec>;
2323

24-
const GPT_TURBO: Spec = {
24+
const GPT_TURBO_128k: Spec = {
2525
display: "OpenAI GPT-4 Turbo 128k",
2626
color: "#10a37f",
2727
category: "ai",
2828
} as const;
2929

3030
const GPT_TURBO_8K: Spec = {
31-
...GPT_TURBO,
32-
display: `${GPT_TURBO.display} 8k`,
31+
...GPT_TURBO_128k,
32+
display: "OpenAI GPT-4 Turbo",
3333
} as const;
3434

35-
const GPT_OMNI: Spec = {
36-
display: "OpenAI GPT-4 Omni",
35+
const GPT_OMNI_128k: Spec = {
36+
display: "OpenAI GPT-4o 128k",
3737
color: "#10a37f",
3838
category: "ai",
3939
} as const;
4040

4141
const GPT_OMNI_8K: Spec = {
42-
...GPT_OMNI,
43-
display: `${GPT_OMNI.display} 8k`,
42+
...GPT_OMNI_128k,
43+
display: "OpenAI GPT-4o",
4444
} as const;
4545

46-
const GPT_OMNI_MINI: Spec = {
47-
...GPT_OMNI,
48-
display: "OpenAI GPT-4o Mini",
46+
const GPT_OMNI_MINI_128k: Spec = {
47+
...GPT_OMNI_128k,
48+
display: "OpenAI GPT-4o Mini 128k",
4949
};
5050

5151
const GPT_OMNI_MINI_8K: Spec = {
52-
...GPT_OMNI_MINI,
53-
display: `${GPT_OMNI_MINI.display} 8k`,
52+
...GPT_OMNI_MINI_128k,
53+
display: "OpenAI GPT-4o Mini",
5454
};
5555

5656
const GOOGLE_AI_COLOR = "#ff4d4f";
@@ -120,13 +120,13 @@ export const QUOTA_SPEC: QuotaSpec = {
120120
color: "#10a37f",
121121
category: "ai",
122122
},
123-
"openai-gpt-4-turbo-preview": GPT_TURBO, // the "preview" is over
123+
"openai-gpt-4-turbo-preview": GPT_TURBO_128k, // the "preview" is over
124124
"openai-gpt-4-turbo-preview-8k": GPT_TURBO_8K, // the "preview" is over
125-
"openai-gpt-4-turbo": GPT_TURBO,
125+
"openai-gpt-4-turbo": GPT_TURBO_128k,
126126
"openai-gpt-4-turbo-8k": GPT_TURBO_8K,
127-
"openai-gpt-4o": GPT_OMNI,
127+
"openai-gpt-4o": GPT_OMNI_128k,
128128
"openai-gpt-4o-8k": GPT_OMNI_8K,
129-
"openai-gpt-4o-mini": GPT_OMNI_MINI,
129+
"openai-gpt-4o-mini": GPT_OMNI_MINI_128k,
130130
"openai-gpt-4o-mini-8k": GPT_OMNI_MINI_8K,
131131
"google-text-bison-001": {
132132
display: "Google Palm 2 (Text)",

0 commit comments

Comments
 (0)