Skip to content

Commit a0bb3cb

Browse files
rekram1-noderekram1-node
committed
fix(vercel): use base_model for new catalog entries and add Thinking Machines
Wire Opus Fast, Realtime Whisper, and Inkling through model metadata, and add a Thinking Machines provider for first-party Inkling access.
1 parent 976ca20 commit a0bb3cb

10 files changed

Lines changed: 89 additions & 68 deletions

File tree

‎models/openai/gpt-realtime-whisper.toml‎

Lines changed: 18 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,18 @@
1+
name = "GPT Realtime Whisper"
2+
description = "Streaming speech-to-text model for low-latency transcript deltas from live audio"
3+
family = "whisper"
4+
release_date = "2026-05-07"
5+
last_updated = "2026-05-07"
6+
attachment = false
7+
reasoning = false
8+
temperature = true
9+
tool_call = false
10+
open_weights = false
11+
12+
[limit]
13+
context = 0
14+
output = 0
15+
16+
[modalities]
17+
input = ["audio"]
18+
output = ["text"]

‎models/thinkingmachines/inkling.toml‎

Lines changed: 30 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,30 @@
1+
# Sources (accessed 2026-07-16):
2+
# - https://thinkingmachines.ai/news/introducing-inkling/
3+
# - https://thinkingmachines.ai/model-card/inkling/
4+
# - https://huggingface.co/thinkingmachines/Inkling
5+
# - https://tinker-docs.thinkingmachines.ai/tinker/models/
6+
# - https://tinker-docs.thinkingmachines.ai/cookbook/inkling/thinking-effort/
7+
# - https://tinker-docs.thinkingmachines.ai/tinker/compatible-apis/openai/
8+
9+
name = "Inkling"
10+
description = "Multimodal MoE reasoning model (975B total, 41B active) for text, image, and audio"
11+
family = "ling"
12+
release_date = "2026-07-15"
13+
last_updated = "2026-07-15"
14+
attachment = true
15+
reasoning = true
16+
temperature = true
17+
tool_call = true
18+
open_weights = true
19+
20+
[limit]
21+
context = 256_000
22+
output = 256_000
23+
24+
[modalities]
25+
input = ["text", "image", "audio", "pdf"]
26+
output = ["text"]
27+
28+
[[weights]]
29+
label = "Hugging Face"
30+
url = "https://huggingface.co/thinkingmachines/Inkling"

‎packages/core/src/sync/providers/openrouter.ts‎

Lines changed: 3 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -15,6 +15,8 @@ const CANONICAL_BASE_MODEL_OVERRIDES = {
1515
"openai/gpt-5.6-luna-pro": "openai/gpt-5.6-luna",
1616
"openai/gpt-5.6-sol-pro": "openai/gpt-5.6-sol",
1717
"openai/gpt-5.6-terra-pro": "openai/gpt-5.6-terra",
18+
"anthropic/claude-opus-4.7-fast": "anthropic/claude-opus-4-7",
19+
"anthropic/claude-opus-4.8-fast": "anthropic/claude-opus-4-8",
1820
} as const;
1921

2022
const CANONICAL_PROVIDER_PREFIXES = {
@@ -33,6 +35,7 @@ const CANONICAL_PROVIDER_PREFIXES = {
3335
qwen: { provider: "alibaba", metadata: "alibaba" },
3436
stepfun: { provider: "stepfun", metadata: "stepfun" },
3537
tencent: { provider: "tencent", metadata: "tencent" },
38+
thinkingmachines: { provider: "thinkingmachines", metadata: "thinkingmachines" },
3639
"x-ai": { provider: "xai", metadata: "xai" },
3740
xai: { provider: "xai", metadata: "xai" },
3841
xiaomi: { provider: "xiaomi", metadata: "xiaomi" },

‎providers/thinkingmachines/logo.svg‎

Lines changed: 3 additions & 0 deletions
Loading

‎providers/thinkingmachines/models/inkling.toml‎

Lines changed: 15 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,15 @@
1+
# Sources (accessed 2026-07-16):
2+
# - Pricing (256K): https://tinker-docs.thinkingmachines.ai/tinker/models/
3+
# - Reasoning: https://tinker-docs.thinkingmachines.ai/tinker/compatible-apis/openai/
4+
# - Effort presets: https://tinker-docs.thinkingmachines.ai/cookbook/inkling/thinking-effort/
5+
# API: {"reasoning_effort": "none"|"minimal"|"low"|"medium"|"high"|"xhigh"} or float [0.0, 0.99]
6+
base_model = "thinkingmachines/inkling"
7+
reasoning_options = [{ type = "effort", values = ["none", "minimal", "low", "medium", "high", "xhigh"] }]
8+
9+
[interleaved]
10+
field = "reasoning_content"
11+
12+
[cost]
13+
input = 3.74
14+
output = 9.36
15+
cache_read = 0.748

‎providers/thinkingmachines/provider.toml‎

Lines changed: 11 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,11 @@
1+
# OpenAI Chat is POST `/chat/completions` with `reasoning_effort` =
2+
# "none"|"minimal"|"low"|"medium"|"high"|"xhigh" or a float in [0.0, 0.99].
3+
# Reasoning traces can be returned on `reasoning_content` via
4+
# `separate_reasoning` (default true).
5+
# https://tinker-docs.thinkingmachines.ai/tinker/compatible-apis/openai/ (accessed 2026-07-16)
6+
# https://tinker-docs.thinkingmachines.ai/cookbook/inkling/thinking-effort/ (accessed 2026-07-16)
7+
name = "Thinking Machines"
8+
env = ["TINKER_API_KEY"]
9+
npm = "@ai-sdk/openai-compatible"
10+
api = "https://tinker.thinkingmachines.dev/services/tinker-prod/oai/api/v1"
11+
doc = "https://tinker-docs.thinkingmachines.ai/tinker/compatible-apis/openai/"

‎providers/vercel/models/anthropic/claude-opus-4.7-fast.toml‎

Lines changed: 2 additions & 17 deletions
Original file line numberDiff line numberDiff line change
@@ -1,25 +1,10 @@
1+
base_model = "anthropic/claude-opus-4-7"
12
name = "Claude Opus 4.7 (Fast)"
2-
description = "Flagship Claude model for deep reasoning, coding, and long-horizon agents"
3-
family = "claude-opus"
4-
release_date = "2026-04-16"
5-
last_updated = "2026-04-16"
6-
attachment = true
7-
reasoning = true
3+
reasoning_options = [{ type = "toggle" }, { type = "effort", values = ["low", "medium", "high", "xhigh"] }]
84
temperature = true
9-
tool_call = true
10-
open_weights = false
11-
reasoning_options = []
125

136
[cost]
147
input = 30
158
output = 150
169
cache_read = 3
1710
cache_write = 37.5
18-
19-
[limit]
20-
context = 1_000_000
21-
output = 128_000
22-
23-
[modalities]
24-
input = ["text", "image", "pdf"]
25-
output = ["text"]

‎providers/vercel/models/anthropic/claude-opus-4.8-fast.toml‎

Lines changed: 2 additions & 17 deletions
Original file line numberDiff line numberDiff line change
@@ -1,25 +1,10 @@
1+
base_model = "anthropic/claude-opus-4-8"
12
name = "Claude Opus 4.8 (Fast)"
2-
description = "Flagship Claude model for deep reasoning, coding, and long-horizon agents"
3-
family = "claude-opus"
4-
release_date = "2026-05-28"
5-
last_updated = "2026-05-28"
6-
attachment = true
7-
reasoning = true
3+
reasoning_options = [{ type = "toggle" }, { type = "effort", values = ["low", "medium", "high", "xhigh"] }]
84
temperature = true
9-
tool_call = true
10-
open_weights = false
11-
reasoning_options = []
125

136
[cost]
147
input = 10
158
output = 50
169
cache_read = 1
1710
cache_write = 12.5
18-
19-
[limit]
20-
context = 1_000_000
21-
output = 128_000
22-
23-
[modalities]
24-
input = ["text", "image", "pdf"]
25-
output = ["text"]

‎providers/vercel/models/openai/gpt-realtime-whisper.toml‎

Lines changed: 1 addition & 17 deletions
Original file line numberDiff line numberDiff line change
@@ -1,18 +1,2 @@
1+
base_model = "openai/gpt-realtime-whisper"
12
name = "gpt-realtime-whisper"
2-
description = "Speech transcription model for accurate audio-to-text and captioning workflows"
3-
family = "whisper"
4-
release_date = "2026-05-07"
5-
last_updated = "2026-05-07"
6-
attachment = false
7-
reasoning = false
8-
temperature = true
9-
tool_call = false
10-
open_weights = false
11-
12-
[limit]
13-
context = 0
14-
output = 0
15-
16-
[modalities]
17-
input = ["audio"]
18-
output = ["text"]

‎providers/vercel/models/thinkingmachines/inkling.toml‎

Lines changed: 4 additions & 17 deletions
Original file line numberDiff line numberDiff line change
@@ -1,24 +1,11 @@
1+
# Sources (accessed 2026-07-16):
2+
# - https://ai-gateway.vercel.sh/v1/models (thinkingmachines/inkling)
3+
base_model = "thinkingmachines/inkling"
14
name = "Inkling"
2-
description = "Multimodal reasoning model for visual analysis, planning, and tool use"
3-
family = "ling"
4-
release_date = "2026-07-15"
5-
last_updated = "2026-07-15"
6-
attachment = true
7-
reasoning = true
8-
temperature = true
9-
tool_call = true
10-
open_weights = false
5+
open_weights = true
116
reasoning_options = []
127

138
[cost]
149
input = 1
1510
output = 4.05
1611
cache_read = 0.17
17-
18-
[limit]
19-
context = 256_000
20-
output = 256_000
21-
22-
[modalities]
23-
input = ["text", "image", "pdf"]
24-
output = ["text"]

0 commit comments

Comments
 (0)