@LLMDevPass

One AI coding subscription. Every model. by @llmgateway

 Earth
devpass.llmgateway.io
Joined August 2026
  • Tweets261
  • Following30
  • Followers123
69Media
Nine days after release, DeepSeek V4.1 Flash 1017 took the week on LLM Gateway: 109.4B tokens, 34.6% of the total, up 340.3%. The original V4.1 Flash fell 45.6% to 86.7B. Fastest riser past 1B tokens: GPT-6 Luna Decisions, +568.9%. 316.4B tokens, 7.7M requests in 7 days.
Stacked bar chart of hourly LLM Gateway token volume for the top 8 models over the 7 days to Oct 26: 316.4B tokens, 7.7M requests, 197 models ranked, led by DeepSeek V4.1 Flash 1017 (34.6%), DeepSeek V4.1 Flash (27.4%) and GPT-6 Luna (9.4%).

ALT Stacked bar chart of hourly LLM Gateway token volume for the top 8 models over the 7 days to Oct 26: 316.4B tokens, 7.7M requests, 197 models ranked, led by DeepSeek V4.1 Flash 1017 (34.6%), DeepSeek V4.1 Flash (27.4%) and GPT-6 Luna (9.4%).

1
41
Luna Decisions is #16 by tokens (2.0B) but #1 by requests: 1.9M calls, 24.8% of the week's total, at ~1.1k tokens each vs 41.3k board-wide. Sort by tokens alone and you'd miss it. Requests per model, live: llmgateway.io/rankings

LLM Rankings — Top Models by Real Usage

Live LLM rankings from real traffic routed through LLM Gateway: top models and apps by token volume, usage trends, and provider market share.

llmgateway.io
2
The fastest grower in LLM Gateway's 500M+ token club isn't on this chart. MiniMax M3.1 Flash, released Oct 19: +27,641.9% in 24h, to 543.1M tokens. #9, just below the cut. Up top, both DeepSeek V4.1 Flash builds shrank (-19.1%, -42.6%) and still took 67.9% of 29.9B tokens.
Stacked bar chart of hourly token volume by model on LLM Gateway over the last 24 hours: 29.9B tokens, 988.3k requests and 139 models ranked, led by DeepSeek V4.1 Flash 1017 (47.0%) and DeepSeek V4.1 Flash (20.9%).

ALT Stacked bar chart of hourly token volume by model on LLM Gateway over the last 24 hours: 29.9B tokens, 988.3k requests and 139 models ranked, led by DeepSeek V4.1 Flash 1017 (47.0%) and DeepSeek V4.1 Flash (20.9%).

1
1
1
102
Following a mover like that takes one model string. All 139 models, ranked live: llmgateway.io/rankings

LLM Rankings — Top Models by Real Usage

Live LLM rankings from real traffic routed through LLM Gateway: top models and apps by token volume, usage trends, and provider market share.

llmgateway.io
11
Stop paying for a coding plan that won't show you what a request cost. DevPass Pro: $79/mo, $158 of usage at provider list rates, per-request cost on a live dashboard. Published limits: 9% of the allowance per rolling 24h, 12% a week on premium models. devpass.llmgateway.io

DevPass by LLM Gateway - All-Access Dev Plans for AI Coding

One subscription, every coding model. Fixed-price dev plans for Claude Code, Cursor, and 100+ models.

devpass.llmgateway.io
21
Eight models moved 500M+ tokens through LLM Gateway in the last 24h. One of them grew. MiMo V2.6 Pro more than doubled: +109.4% to 763.6M tokens, on just 2,014 requests. The other seven fell, from -4.0% (DeepSeek V4 Pro) to -66.7% (GPT-6 Luna). 38.9B tokens, 170 models.
LLM Gateway rankings for the last 24 hours: 38.9B tokens processed, 1M requests routed and 170 models ranked, with an hourly stacked bar chart of the top 8 models led by DeepSeek V4.1 Flash 1017 (44.9%), DeepSeek V4.1 Flash (28.2%) and GPT-6 Luna (6.0%).

ALT LLM Gateway rankings for the last 24 hours: 38.9B tokens processed, 1M requests routed and 170 models ranked, with an hourly stacked bar chart of the top 8 models led by DeepSeek V4.1 Flash 1017 (44.9%), DeepSeek V4.1 Flash (28.2%) and GPT-6 Luna (6.0%).

1
1
1
76
Seven down, one up in the top 8. All 170 models, ranked live: llmgateway.io/rankings

LLM Rankings — Top Models by Real Usage

Live LLM rankings from real traffic routed through LLM Gateway: top models and apps by token volume, usage trends, and provider market share.

llmgateway.io
12
Kimi K3.1 Pro and K3.1 Pro Highspeed are live on LLM Gateway: Moonshot's flagship reasoning models. 1M context, vision + tools, always-on thinking. Pro: $2/M in, $10/M out Highspeed: $4/M in, $20/M out llmgateway.io/models/kimi-k3…

Kimi K3.1 Pro — Pricing, Providers & Benchmarks

Kimi's flagship reasoning model with always-on thinking, vision input, tool calling, and a 1M-token context window. Use Kimi K3.1 Pro via one OpenAI-compatible API with live per-token pricing across...

llmgateway.io
1
3
154
The most-called model on LLM Gateway in the last 24h doesn't write a word. GPT-6 Luna Decisions: 260.0k requests, 24.0% of all 1.1M, on 288.2M tokens. ~1.1k tokens a call vs 51.7k board-wide. By tokens it ranks #14. Two DeepSeek V4.1 Flash builds took 64.5% of 55.9B.
Stacked bar chart of hourly token volume by model on LLM Gateway over the last 24 hours, with 55.9B tokens, 1.1M requests and 146 models ranked, led by DeepSeek V4.1 Flash 1017 at 37.6%, DeepSeek V4.1 Flash at 26.9% and GPT-6 Luna at 12.8%.

ALT Stacked bar chart of hourly token volume by model on LLM Gateway over the last 24 hours, with 55.9B tokens, 1.1M requests and 146 models ranked, led by DeepSeek V4.1 Flash 1017 at 37.6%, DeepSeek V4.1 Flash at 26.9% and GPT-6 Luna at 12.8%.

1
1
1
84
The chart stops at 8. Where the other 138 landed, live: llmgateway.io/rankings

LLM Rankings — Top Models by Real Usage

Live LLM rankings from real traffic routed through LLM Gateway: top models and apps by token volume, usage trends, and provider market share.

llmgateway.io
12
Hot take: a coding plan locked to one lab's models is a bet that lab stays on top all year. Rankings don't sit still. DevPass: one key for every model in its catalogue, in Claude Code, Cursor or Cline. $29 → $58 of usage at provider list rates. devpass.llmgateway.io

DevPass by LLM Gateway - All-Access Dev Plans for AI Coding

One subscription, every coding model. Fixed-price dev plans for Claude Code, Cursor, and 100+ models.

devpass.llmgateway.io
34
Kimi K3 launched in July. In the last 24h its token volume on LLM Gateway jumped 1022.6%, to 1.3B. Also up: GPT-6.1 Sol +113.1% to 2.9B, Claude Opus 5.5 +93.7% to 2.0B. At #1 and #2, the two DeepSeek V4.1 Flash builds finished 35.9M tokens apart: 28.5% vs 28.4% of 56.3B.
Stacked bar chart of hourly token volume on LLM Gateway over the last 24 hours (56.3B tokens, 1.1M requests, 144 models ranked), led by DeepSeek V4.1 Flash 1017 at 28.5%, DeepSeek V4.1 Flash at 28.4% and GPT-6 Luna at 13.0%.

ALT Stacked bar chart of hourly token volume on LLM Gateway over the last 24 hours (56.3B tokens, 1.1M requests, 144 models ranked), led by DeepSeek V4.1 Flash 1017 at 28.5%, DeepSeek V4.1 Flash at 28.4% and GPT-6 Luna at 13.0%.

1
55
A July model at #8 and a photo finish at #1. All 144, ranked live: llmgateway.io/rankings

LLM Rankings — Top Models by Real Usage

Live LLM rankings from real traffic routed through LLM Gateway: top models and apps by token volume, usage trends, and provider market share.

llmgateway.io
9
DeepSeek lost #1 on LLM Gateway. To DeepSeek. V4.1 Flash 1017, released Oct 17, moved 15.9B tokens in 24h (29.9% of the board), up 30.5%. The original slipped 3.4% to 13.7B. Biggest jump: MiMo V2.6 Flash, +704.6% to 884.0M. 53.0B tokens, 1.4M requests, 140 models.
Stacked bar chart of hourly token volume through LLM Gateway over the last 24 hours, showing 53B tokens, 1.4M requests and 140 models ranked, led by DeepSeek V4.1 Flash 1017 at 29.9%, DeepSeek V4.1 Flash at 26.2% and GPT-6 Luna at 12.5%.

ALT Stacked bar chart of hourly token volume through LLM Gateway over the last 24 hours, showing 53B tokens, 1.4M requests and 140 models ranked, led by DeepSeek V4.1 Flash 1017 at 29.9%, DeepSeek V4.1 Flash at 26.2% and GPT-6 Luna at 12.5%.

1
1
1
62
Hour-by-hour split of the top 8, plus the 132 ranked below them: llmgateway.io/rankings

LLM Rankings — Top Models by Real Usage

Live LLM rankings from real traffic routed through LLM Gateway: top models and apps by token volume, usage trends, and provider market share.

llmgateway.io
5
"Fair use" shouldn't be a riddle. DevPass caps, in dollars: Lite $29/mo: $58 of usage, up to $4.64 per rolling 24h Pro $79: $158, up to $14.22 Max $179: $358, up to $35.80 Premium models: 10-15% of credits a week. Provider list rates, no token markup. devpass.llmgateway.io

DevPass by LLM Gateway - All-Access Dev Plans for AI Coding

One subscription, every coding model. Fixed-price dev plans for Claude Code, Cursor, and 100+ models.

devpass.llmgateway.io
32
Both DeepSeek snapshots atop LLM Gateway lost volume in 24h: V4.1 Flash -7.5%, the 1017 build -11.7%. #3 and #4 more than doubled: GPT-6 Luna +132.2% to 3.8B tokens, GLM-5.3 +162.4% to 3.7B. 47.6B tokens, 990.9k requests, 142 models. Hard-coded model choices age fast.
LLM Gateway rankings for the last 24 hours: 47.6B tokens processed, 990.9k requests routed and 142 models ranked, with an hourly stacked bar chart of the top 8 models led by DeepSeek V4.1 Flash (30.4%), DeepSeek V4.1 Flash 1017 (25.8%) and GPT-6 Luna (8.0%).

ALT LLM Gateway rankings for the last 24 hours: 47.6B tokens processed, 990.9k requests routed and 142 models ranked, with an hourly stacked bar chart of the top 8 models led by DeepSeek V4.1 Flash (30.4%), DeepSeek V4.1 Flash 1017 (25.8%) and GPT-6 Luna (8.0%).

1
1
1
79
All 142 on the live board. See if the top two claw it back: llmgateway.io/rankings

LLM Rankings — Top Models by Real Usage

Live LLM rankings from real traffic routed through LLM Gateway: top models and apps by token volume, usage trends, and provider market share.

llmgateway.io
5
More than half of the week's tokens went to one model. 305.2B tokens and 6.6M requests through LLM Gateway in 7 days, across 221 models. DeepSeek V4.1 Flash took 160.1B of them (52.5%), up 50.0% on the week before. Biggest climber: MiMo V2.6 Flash, up 237.3% to 6.0B.
LLM Gateway rankings for the last 7 days: 305.2B tokens processed, 6.6M requests routed, 221 models ranked, with an hourly stacked bar chart of token volume led by DeepSeek V4.1 Flash (52.5%), DeepSeek V4.1 Flash 1017 (8.3%) and GLM-5.3 Flash (7.4%).

ALT LLM Gateway rankings for the last 7 days: 305.2B tokens processed, 6.6M requests routed, 221 models ranked, with an hourly stacked bar chart of token volume led by DeepSeek V4.1 Flash (52.5%), DeepSeek V4.1 Flash 1017 (8.3%) and GLM-5.3 Flash (7.4%).

1
1
2
100
Behind the leader it's churn: 116 of the 221 moved fewer tokens than the prior week, and #2 is a DeepSeek snapshot that only went live Oct 17 (25.2B tokens). Hard-code one model and you're betting against the board. Watch it move: llmgateway.io/rankings

LLM Rankings — Top Models by Real Usage

Live LLM rankings from real traffic routed through LLM Gateway: top models and apps by token volume, usage trends, and provider market share.

llmgateway.io
25
Load more