@grida/ai-models 1.20261112.3 → 1.20261113.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/PROVENANCE.md +48 -0
- package/README.md +2 -2
- package/manifest.json +10 -10
- package/models.json +197 -0
- package/package.json +1 -1
package/PROVENANCE.md
CHANGED
|
@@ -1844,3 +1844,51 @@ published component uses one-unit quantities with no additional quantity
|
|
|
1844
1844
|
rounding. TTS and sound-effects settings retain their distinct request schemas;
|
|
1845
1845
|
these Grida bindings do not imply support for every native voice, format or
|
|
1846
1846
|
operation.
|
|
1847
|
+
|
|
1848
|
+
## `google/gemini-4-hydrogen`
|
|
1849
|
+
|
|
1850
|
+
Google's November 12, 2026 announcement establishes broad public availability of
|
|
1851
|
+
the exact model in the Gemini API and AI Studio. The native model accepts text,
|
|
1852
|
+
images, audio, video and PDF and returns text.
|
|
1853
|
+
https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-hydrogen/
|
|
1854
|
+
https://aistudio.google.com/docs/models/gemini-4-hydrogen
|
|
1855
|
+
|
|
1856
|
+
Regular paid Standard prices are independently sourced for Google, Vercel and
|
|
1857
|
+
OpenRouter: $1.50 input, $7.50 output and $0.075 implicit-cache read per million
|
|
1858
|
+
tokens. Native output includes thinking. Vercel's exact Google and Google Vertex
|
|
1859
|
+
endpoint records publish those decimal per-token rates with no discount; the
|
|
1860
|
+
page's rounded $0.08 cache display is not the evidence for $0.075. OpenRouter's
|
|
1861
|
+
Standard provider rows independently publish the exact rates. Optional tools,
|
|
1862
|
+
alternate service tiers and regional premiums are outside these token schedules.
|
|
1863
|
+
https://aistudio.google.com/docs/pricing
|
|
1864
|
+
https://ai-gateway.vercel.sh/v1/models/google/gemini-4-hydrogen/endpoints
|
|
1865
|
+
https://openrouter.ai/google/gemini-4-hydrogen
|
|
1866
|
+
|
|
1867
|
+
The native 1,048,576-token window includes input, thinking and output. The
|
|
1868
|
+
intrinsic/native output field records the 65,536-token foreground segment limit
|
|
1869
|
+
for streaming and non-streaming. Longer output requires background execution or
|
|
1870
|
+
continuation. Vercel's 1,000,000-token output ceiling and OpenRouter's 943,718-token
|
|
1871
|
+
ceiling remain provider discovery facts, not synchronous-call guarantees.
|
|
1872
|
+
https://aistudio.google.com/docs/long-decode
|
|
1873
|
+
https://ai-gateway.vercel.sh/v1/models
|
|
1874
|
+
https://openrouter.ai/api/v1/models
|
|
1875
|
+
|
|
1876
|
+
Native thinking supports low, medium, high and xhigh, defaulting to high. MINIMAL
|
|
1877
|
+
and thinking_budget are invalid, and custom sampling controls are ignored.
|
|
1878
|
+
OpenRouter instead lists mandatory reasoning with medium as its provider default;
|
|
1879
|
+
Vercel discovery lists the four efforts without establishing omitted-effort
|
|
1880
|
+
mapping for every compatible API. Current native caching is implicit only with a
|
|
1881
|
+
4,096-token threshold. OpenRouter's cache-write amount lacks a verified current
|
|
1882
|
+
operation/TTL/unit and is left unpriced; omission is not a zero price. Forward
|
|
1883
|
+
cache-control documentation does not establish current cache-write behavior.
|
|
1884
|
+
https://aistudio.google.com/docs/whats-new-gemini-4-hydrogen
|
|
1885
|
+
https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/4-hydrogen
|
|
1886
|
+
https://ai-gateway.vercel.sh/v1/models/google/gemini-4-hydrogen/endpoints
|
|
1887
|
+
https://openrouter.ai/api/v1/models
|
|
1888
|
+
|
|
1889
|
+
Google's media-resolution table explicitly covers Gemini 4 and Gemini 3, with
|
|
1890
|
+
Hydrogen examples. Image budgets are approximate: unspecified/high 1,120, low 280,
|
|
1891
|
+
medium 560 and ultra_high 2,240. This maximum comes from the allocation table, not
|
|
1892
|
+
the model page's default resolution. Image MIME types are sourced separately.
|
|
1893
|
+
https://aistudio.google.com/docs/media-resolution
|
|
1894
|
+
https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/4-hydrogen
|
package/README.md
CHANGED
|
@@ -5,7 +5,7 @@ provider bindings, and published prices, with the labs that make them and their
|
|
|
5
5
|
logos. Every fact cites a first-party source.
|
|
6
6
|
|
|
7
7
|
This package is the data, published unchanged from release
|
|
8
|
-
[`v1.
|
|
8
|
+
[`v1.20261113.0`](https://github.com/gridaco/models/releases/tag/v1.20261113.0) of [gridaco/models](https://github.com/gridaco/models). It contains no code.
|
|
9
9
|
|
|
10
10
|
| file | contents |
|
|
11
11
|
| --- | --- |
|
|
@@ -25,7 +25,7 @@ From a CDN, without installing:
|
|
|
25
25
|
|
|
26
26
|
```
|
|
27
27
|
https://cdn.jsdelivr.net/npm/@grida/ai-models@1/models.json
|
|
28
|
-
https://cdn.jsdelivr.net/npm/@grida/ai-models@1.
|
|
28
|
+
https://cdn.jsdelivr.net/npm/@grida/ai-models@1.20261113.0/models.json
|
|
29
29
|
```
|
|
30
30
|
|
|
31
31
|
## Versions
|
package/manifest.json
CHANGED
|
@@ -1,19 +1,19 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "grida-models",
|
|
3
|
-
"version": "v1.
|
|
3
|
+
"version": "v1.20261113.0",
|
|
4
4
|
"schema_version": 1,
|
|
5
|
-
"commit": "
|
|
6
|
-
"created": "2026-11-
|
|
7
|
-
"source": "https://github.com/gridaco/models/tree/
|
|
8
|
-
"digest": "sha256:
|
|
5
|
+
"commit": "22659ac127a10b56a87bd541020c7ee8e1f75091",
|
|
6
|
+
"created": "2026-11-13T09:32:26Z",
|
|
7
|
+
"source": "https://github.com/gridaco/models/tree/22659ac127a10b56a87bd541020c7ee8e1f75091/catalogue",
|
|
8
|
+
"digest": "sha256:4cc3236c84546e3fd114d6b2b4143d281c0238d859dde46bf680f7c74112f24e",
|
|
9
9
|
"counts": {
|
|
10
|
-
"cards":
|
|
10
|
+
"cards": 106,
|
|
11
11
|
"labs": 18
|
|
12
12
|
},
|
|
13
13
|
"files": {
|
|
14
14
|
"PROVENANCE.md": {
|
|
15
|
-
"sha256": "
|
|
16
|
-
"bytes":
|
|
15
|
+
"sha256": "4be6aa57f189a5072de0396227f1f68b2b1ee4029e2cc96aa44e71c1d99a7f03",
|
|
16
|
+
"bytes": 101464
|
|
17
17
|
},
|
|
18
18
|
"README.md": {
|
|
19
19
|
"sha256": "c4251f78423708a1c6cfbe0cfa4c208fb7cecbfc14ddd512bb46c5671abeae7a",
|
|
@@ -92,8 +92,8 @@
|
|
|
92
92
|
"bytes": 330
|
|
93
93
|
},
|
|
94
94
|
"models.json": {
|
|
95
|
-
"sha256": "
|
|
96
|
-
"bytes":
|
|
95
|
+
"sha256": "156ce2cd0a3210525a7ccc58fd847a9c53c2988f3d7f4dec3960f89cbea2f94f",
|
|
96
|
+
"bytes": 1121561
|
|
97
97
|
},
|
|
98
98
|
"schemas/labs.schema.json": {
|
|
99
99
|
"sha256": "9ddaedd56826bebdbe72b84f76cfc86df0b5c5902ba9f0259c2ea2cbb9327f99",
|
package/models.json
CHANGED
|
@@ -3598,6 +3598,203 @@
|
|
|
3598
3598
|
}
|
|
3599
3599
|
}
|
|
3600
3600
|
},
|
|
3601
|
+
"google/gemini-4-hydrogen": {
|
|
3602
|
+
"id": "google/gemini-4-hydrogen",
|
|
3603
|
+
"vendor": "google",
|
|
3604
|
+
"reference_provider": "google",
|
|
3605
|
+
"label": "Gemini 4 Hydrogen",
|
|
3606
|
+
"multimodal": true,
|
|
3607
|
+
"image_input_mimes": [
|
|
3608
|
+
"image/png",
|
|
3609
|
+
"image/jpeg",
|
|
3610
|
+
"image/webp",
|
|
3611
|
+
"image/heic",
|
|
3612
|
+
"image/heif"
|
|
3613
|
+
],
|
|
3614
|
+
"tool_call": true,
|
|
3615
|
+
"context_window": 1048576,
|
|
3616
|
+
"output_limit": 65536,
|
|
3617
|
+
"cost": {
|
|
3618
|
+
"input": 1.5,
|
|
3619
|
+
"output": 7.5,
|
|
3620
|
+
"cache_read": 0.075
|
|
3621
|
+
},
|
|
3622
|
+
"release": {
|
|
3623
|
+
"date": "2026-11-12",
|
|
3624
|
+
"basis": "model",
|
|
3625
|
+
"source_url": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-hydrogen/"
|
|
3626
|
+
},
|
|
3627
|
+
"vision_tokenization": {
|
|
3628
|
+
"source_url": "https://aistudio.google.com/docs/media-resolution",
|
|
3629
|
+
"maximum_tokens_per_image": 2240,
|
|
3630
|
+
"token_counts_approximate": true,
|
|
3631
|
+
"detail_token_budgets": {
|
|
3632
|
+
"unspecified": 1120,
|
|
3633
|
+
"low": 280,
|
|
3634
|
+
"medium": 560,
|
|
3635
|
+
"high": 1120,
|
|
3636
|
+
"ultra_high": 2240
|
|
3637
|
+
}
|
|
3638
|
+
},
|
|
3639
|
+
"providers": {
|
|
3640
|
+
"google": {
|
|
3641
|
+
"provider": "google",
|
|
3642
|
+
"id": "gemini-4-hydrogen",
|
|
3643
|
+
"url": "https://aistudio.google.com/docs/models/gemini-4-hydrogen",
|
|
3644
|
+
"source_url": "https://aistudio.google.com/docs/models/gemini-4-hydrogen",
|
|
3645
|
+
"endpoints": [
|
|
3646
|
+
{
|
|
3647
|
+
"id": "gemini-4-hydrogen",
|
|
3648
|
+
"input": "text",
|
|
3649
|
+
"source_url": "https://aistudio.google.com/docs/models/gemini-4-hydrogen"
|
|
3650
|
+
}
|
|
3651
|
+
],
|
|
3652
|
+
"qualifications": [
|
|
3653
|
+
{
|
|
3654
|
+
"source_url": "https://aistudio.google.com/docs/long-decode",
|
|
3655
|
+
"notes": "One foreground segment permits 65,536 output tokens with or without streaming, including thinking. Longer output requires continuation or background execution within the unified 1,048,576-token input/thinking/output window."
|
|
3656
|
+
},
|
|
3657
|
+
{
|
|
3658
|
+
"source_url": "https://aistudio.google.com/docs/whats-new-gemini-4-hydrogen",
|
|
3659
|
+
"notes": "Text, images, audio, video and PDF inputs produce text. Thinking levels are low, medium, high and xhigh, with high as the native default."
|
|
3660
|
+
},
|
|
3661
|
+
{
|
|
3662
|
+
"source_url": "https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/4-hydrogen",
|
|
3663
|
+
"notes": "MINIMAL thinking and thinking_budget are invalid; custom temperature, top_p and top_k are ignored. Current caching is implicit only, with a 4,096-token minimum; explicit caching is unsupported. Image MIME types are image/png, image/jpeg, image/webp, image/heic and image/heif."
|
|
3664
|
+
}
|
|
3665
|
+
],
|
|
3666
|
+
"limits": {
|
|
3667
|
+
"source_url": "https://aistudio.google.com/docs/models/gemini-4-hydrogen",
|
|
3668
|
+
"context_tokens": 1048576,
|
|
3669
|
+
"max_output_tokens": 65536
|
|
3670
|
+
},
|
|
3671
|
+
"capabilities": {
|
|
3672
|
+
"source_url": "https://aistudio.google.com/docs/models/gemini-4-hydrogen",
|
|
3673
|
+
"image_input": true,
|
|
3674
|
+
"tools": true,
|
|
3675
|
+
"reasoning": true,
|
|
3676
|
+
"structured_output": true
|
|
3677
|
+
},
|
|
3678
|
+
"token_pricing": {
|
|
3679
|
+
"source_url": "https://aistudio.google.com/docs/pricing",
|
|
3680
|
+
"unit": "usd_per_token",
|
|
3681
|
+
"scope": "Regular paid Standard token rates. Output includes thinking. Implicit caching only; optional tools, Batch, Flex, Priority and Contributor tiers are excluded.",
|
|
3682
|
+
"input": "0.0000015",
|
|
3683
|
+
"output": "0.0000075",
|
|
3684
|
+
"input_cache_read": "0.000000075"
|
|
3685
|
+
}
|
|
3686
|
+
},
|
|
3687
|
+
"vercel": {
|
|
3688
|
+
"provider": "vercel",
|
|
3689
|
+
"id": "google/gemini-4-hydrogen",
|
|
3690
|
+
"url": "https://vercel.com/ai-gateway/models/gemini-4-hydrogen",
|
|
3691
|
+
"source_url": "https://vercel.com/ai-gateway/models/gemini-4-hydrogen",
|
|
3692
|
+
"endpoints": [
|
|
3693
|
+
{
|
|
3694
|
+
"id": "google/gemini-4-hydrogen",
|
|
3695
|
+
"input": "text",
|
|
3696
|
+
"source_url": "https://vercel.com/ai-gateway/models/gemini-4-hydrogen"
|
|
3697
|
+
}
|
|
3698
|
+
],
|
|
3699
|
+
"qualifications": [
|
|
3700
|
+
{
|
|
3701
|
+
"source_url": "https://aistudio.google.com/docs/long-decode",
|
|
3702
|
+
"notes": "The native model permits 65,536 output tokens per foreground segment, with or without streaming. Longer output uses continuation or background execution; this provider's 1,000,000-token discovery ceiling is not a single synchronous-call guarantee."
|
|
3703
|
+
},
|
|
3704
|
+
{
|
|
3705
|
+
"source_url": "https://ai-gateway.vercel.sh/v1/models/google/gemini-4-hydrogen/endpoints",
|
|
3706
|
+
"notes": "Discovery lists low, medium, high and xhigh reasoning, but does not establish the omitted-effort mapping or default for each compatible API. It lists temperature even though native documentation says custom sampling controls are ignored."
|
|
3707
|
+
}
|
|
3708
|
+
],
|
|
3709
|
+
"limits": {
|
|
3710
|
+
"source_url": "https://ai-gateway.vercel.sh/v1/models",
|
|
3711
|
+
"context_tokens": 1000000,
|
|
3712
|
+
"max_output_tokens": 1000000
|
|
3713
|
+
},
|
|
3714
|
+
"capabilities": {
|
|
3715
|
+
"source_url": "https://ai-gateway.vercel.sh/v1/models",
|
|
3716
|
+
"image_input": true,
|
|
3717
|
+
"tools": true,
|
|
3718
|
+
"reasoning": true,
|
|
3719
|
+
"structured_output": true
|
|
3720
|
+
},
|
|
3721
|
+
"token_pricing": {
|
|
3722
|
+
"source_url": "https://ai-gateway.vercel.sh/v1/models/google/gemini-4-hydrogen/endpoints",
|
|
3723
|
+
"unit": "usd_per_token",
|
|
3724
|
+
"scope": "Regular Standard global token rates from the provider's exact Google and Google Vertex endpoint records, with no discount. Implicit caching only; optional tools, alternate service tiers and regional premiums are excluded.",
|
|
3725
|
+
"input": "0.0000015",
|
|
3726
|
+
"output": "0.0000075",
|
|
3727
|
+
"input_cache_read": "0.000000075"
|
|
3728
|
+
}
|
|
3729
|
+
},
|
|
3730
|
+
"openrouter": {
|
|
3731
|
+
"provider": "openrouter",
|
|
3732
|
+
"id": "google/gemini-4-hydrogen",
|
|
3733
|
+
"url": "https://openrouter.ai/google/gemini-4-hydrogen",
|
|
3734
|
+
"source_url": "https://openrouter.ai/google/gemini-4-hydrogen",
|
|
3735
|
+
"endpoints": [
|
|
3736
|
+
{
|
|
3737
|
+
"id": "google/gemini-4-hydrogen",
|
|
3738
|
+
"input": "text",
|
|
3739
|
+
"source_url": "https://openrouter.ai/google/gemini-4-hydrogen"
|
|
3740
|
+
}
|
|
3741
|
+
],
|
|
3742
|
+
"qualifications": [
|
|
3743
|
+
{
|
|
3744
|
+
"source_url": "https://openrouter.ai/api/v1/models",
|
|
3745
|
+
"notes": "The top_provider.max_completion_tokens discovery limit is 943,718. Reasoning is mandatory; low, medium, high and xhigh are listed, with medium as this provider's default, distinct from the native high default."
|
|
3746
|
+
},
|
|
3747
|
+
{
|
|
3748
|
+
"source_url": "https://aistudio.google.com/docs/long-decode",
|
|
3749
|
+
"notes": "The native model permits 65,536 output tokens per foreground segment, with or without streaming. Longer output uses continuation or background execution; the provider discovery ceiling is not a single synchronous-call guarantee."
|
|
3750
|
+
},
|
|
3751
|
+
{
|
|
3752
|
+
"source_url": "https://openrouter.ai/api/v1/models",
|
|
3753
|
+
"notes": "An input_cache_write amount is present, but its current applicable caching operation, TTL and billing unit are not established for this implicit-only model. No per-token cache-write price is recorded; omission does not mean zero cost. Listed sampling controls do not establish an effect when the native model ignores them."
|
|
3754
|
+
}
|
|
3755
|
+
],
|
|
3756
|
+
"limits": {
|
|
3757
|
+
"source_url": "https://openrouter.ai/api/v1/models",
|
|
3758
|
+
"context_tokens": 1048576,
|
|
3759
|
+
"max_output_tokens": 943718
|
|
3760
|
+
},
|
|
3761
|
+
"capabilities": {
|
|
3762
|
+
"source_url": "https://openrouter.ai/api/v1/models",
|
|
3763
|
+
"image_input": true,
|
|
3764
|
+
"tools": true,
|
|
3765
|
+
"reasoning": true,
|
|
3766
|
+
"structured_output": true
|
|
3767
|
+
},
|
|
3768
|
+
"token_pricing": {
|
|
3769
|
+
"source_url": "https://openrouter.ai/google/gemini-4-hydrogen",
|
|
3770
|
+
"unit": "usd_per_token",
|
|
3771
|
+
"scope": "Regular Standard global token rates explicitly shown for Google Vertex and Google AI Studio. Optional tools, Batch, Flex, Priority and regional premiums are excluded.",
|
|
3772
|
+
"input": "0.0000015",
|
|
3773
|
+
"output": "0.0000075",
|
|
3774
|
+
"input_cache_read": "0.000000075"
|
|
3775
|
+
}
|
|
3776
|
+
}
|
|
3777
|
+
},
|
|
3778
|
+
"provider_coverage": {
|
|
3779
|
+
"google": {
|
|
3780
|
+
"disposition": "listed",
|
|
3781
|
+
"source_url": "https://aistudio.google.com/docs/models/gemini-4-hydrogen"
|
|
3782
|
+
},
|
|
3783
|
+
"vercel": {
|
|
3784
|
+
"disposition": "listed",
|
|
3785
|
+
"source_url": "https://vercel.com/ai-gateway/models/gemini-4-hydrogen"
|
|
3786
|
+
},
|
|
3787
|
+
"openrouter": {
|
|
3788
|
+
"disposition": "listed",
|
|
3789
|
+
"source_url": "https://openrouter.ai/google/gemini-4-hydrogen"
|
|
3790
|
+
},
|
|
3791
|
+
"fal": {
|
|
3792
|
+
"disposition": "excluded",
|
|
3793
|
+
"source_url": "https://fal.ai/models",
|
|
3794
|
+
"reason": "The provider coverage policy includes fal for media models only; no text API compatibility is claimed."
|
|
3795
|
+
}
|
|
3796
|
+
}
|
|
3797
|
+
},
|
|
3601
3798
|
"openai/gpt-5.5": {
|
|
3602
3799
|
"id": "openai/gpt-5.5",
|
|
3603
3800
|
"vendor": "openai",
|