How fast the platform has grown, and how tokens & spend split across vendors.
Clean market (tencent & xiaomi excluded) · Jan 2025 → 2026-07-16 · usage from OpenRouter datasets API, prices from our hourly panel + Wayback. Charts are interactive — hover any point for its value. The final month (Jul) is projected to a full month (÷16×31) so the trend doesn't dip artificially.
760T
cumulative tokens (clean)
$1,605M
cumulative est. spend (clean)
79×
token run-rate (5.6T vs 0.07T/day)
61×
spend run-rate ($13.9M vs $0.23M/day)
29×
quarterly tokens (2026Q2 vs 2025Q1)
26×
quarterly spend (2026Q2 vs 2025Q1)
① Market growth — totals
② Per-vendor dynamics
· or click a name in any legend to toggle it, double-click to isolate one · hover for values
Top 8 vendors by all-time tokens. Absolute-token and absolute-spend charts use a log y-axis because Anthropic dwarfs everyone else.
③ Token share % by month
vendor
25-01
25-02
25-03
25-04
25-05
25-06
25-07
25-08
25-09
25-10
25-11
25-12
26-01
26-02
26-03
26-04
26-05
26-06
26-07*
anthropic
49.3
39.8
35.5
25.1
21.0
21.9
23.4
21.1
14.0
16.2
12.8
14.1
17.8
14.0
15.4
15.5
17.9
17.1
18.3
google
16.8
29.4
32.5
33.2
32.6
39.4
34.2
24.2
18.3
18.5
18.5
22.6
25.3
18.2
16.7
15.2
15.5
11.1
10.0
deepseek
4.3
7.1
10.8
12.2
10.9
15.5
18.2
15.2
10.5
6.8
5.1
7.7
8.9
7.3
7.1
6.9
17.8
20.8
21.0
openai
4.7
4.6
4.3
9.2
21.7
9.9
5.4
8.7
10.6
10.5
8.1
11.7
11.6
10.1
10.5
9.9
9.0
7.6
8.3
minimax
0.4
0.5
0.2
0.1
0.0
0.0
—
—
—
0.7
4.4
2.3
2.2
13.5
11.2
8.7
3.6
10.9
10.7
x-ai
0.2
0.0
0.0
0.3
0.6
1.2
1.4
4.5
29.9
29.8
32.1
18.0
12.8
6.6
4.1
3.6
1.4
0.1
0.6
z-ai
—
—
—
—
—
—
0.2
2.6
1.5
2.9
2.4
2.1
2.6
5.7
5.7
5.4
3.4
4.7
9.1
Qwen
1.3
1.2
1.4
1.2
1.5
0.8
3.7
8.7
3.3
3.6
3.1
3.1
2.2
1.6
2.7
10.9
3.0
1.7
1.1
moonshotai
—
—
—
—
—
—
1.4
2.2
0.6
0.3
0.5
0.4
1.6
8.5
3.5
5.0
5.3
2.0
1.5
openrouter
—
—
—
5.4
—
—
0.2
2.1
2.8
0.1
2.3
0.1
—
0.4
2.7
0.8
4.0
7.2
—
stepfun
—
—
—
—
—
—
—
—
—
—
—
—
—
2.2
7.5
3.4
2.2
2.9
3.0
nvidia
0.1
0.1
—
0.0
—
—
—
—
—
—
0.0
0.5
0.0
0.1
2.0
2.8
2.5
2.9
5.9
other
3.9
4.3
3.5
4.1
5.2
5.5
5.3
5.9
5.5
6.5
6.7
9.2
8.5
6.5
7.9
9.7
11.4
7.8
8.0
market size
2.2T
3.5T
6.3T
8.3T
9.9T
9.7T
11.8T
15.0T
21.4T
22.3T
26.6T
25.3T
29.9T
49.5T
75.2T
90.6T
105.5T
155.3T
177.4T
Token share = vendor's share of monthly platform tokens among the clean market (tencent & xiaomi excluded). Bottom row = total clean-market tokens that month. *26-07 is a partial month (through Jul 16) projected to a full month (÷16×31); share-% and cumulative are unaffected.
④ Est. spend share % by month
vendor
25-01
25-02
25-03
25-04
25-05
25-06
25-07
25-08
25-09
25-10
25-11
25-12
26-01
26-02
26-03
26-04
26-05
26-06
26-07*
anthropic
91.2
89.5
90.3
80.4
69.9
69.4
70.7
67.3
61.6
60.8
54.5
55.5
61.2
58.8
59.6
56.0
60.3
63.1
63.2
google
0.8
1.6
1.9
8.2
14.8
15.9
14.5
11.0
11.4
12.0
16.1
17.8
16.6
12.7
11.1
8.9
7.7
5.6
4.2
deepseek
0.5
1.1
1.7
1.2
1.2
1.5
1.4
1.1
1.3
1.4
1.5
1.5
1.3
1.2
1.1
1.0
2.1
2.3
2.1
openai
1.8
2.4
1.6
4.8
7.8
5.3
3.9
5.9
9.0
7.3
7.4
8.7
7.9
7.9
10.3
10.7
13.3
14.0
15.4
minimax
0.1
0.1
0.0
0.0
0.0
0.0
—
—
—
0.0
0.6
0.6
0.5
3.9
3.1
1.9
0.8
2.4
2.2
x-ai
0.3
0.0
0.0
0.3
0.5
2.1
2.9
4.3
10.0
9.9
10.6
4.6
2.6
1.2
0.6
0.5
0.2
0.0
0.6
z-ai
—
—
—
—
—
—
0.0
0.5
0.4
1.1
1.0
0.9
0.8
2.4
4.0
3.5
1.3
2.8
4.4
Qwen
0.1
0.1
0.1
0.1
0.1
0.1
0.6
1.2
0.5
0.7
0.8
0.9
0.4
0.3
0.4
3.1
0.8
0.9
0.6
moonshotai
—
—
—
—
—
—
0.3
0.4
0.3
0.1
0.3
0.2
0.7
4.5
1.7
3.2
3.0
1.1
0.8
openrouter
—
—
—
0.0
—
—
0.2
2.2
0.0
0.1
0.2
0.1
—
0.4
0.0
0.8
0.0
0.0
—
stepfun
—
—
—
—
—
—
—
—
—
—
—
—
—
0.0
0.0
0.1
0.1
0.5
0.5
nvidia
0.0
0.0
—
0.0
—
—
—
—
—
—
0.0
0.0
0.0
0.0
0.2
0.3
0.2
0.7
0.0
other
3.9
4.4
3.5
4.2
5.1
5.5
5.3
5.9
5.5
6.5
6.7
9.0
8.0
6.5
7.7
9.7
9.9
6.5
6.1
market size
2.2T
3.5T
6.3T
8.3T
9.9T
9.7T
11.8T
15.0T
21.4T
22.3T
26.6T
25.3T
29.9T
49.5T
75.2T
90.6T
105.5T
155.3T
177.4T
Dollar share = tokens × blended list price (0.75·in + 0.25·out). Pre-2026-06 prices are as-of Wayback snapshots (sparse); after that our hourly panel. List-price GMV, not net revenue. *26-07 is a partial month (through Jul 16) projected to a full month (÷16×31); share-% and cumulative are unaffected.
⑤ Price by intelligence (Elo) over time
Each point = one model: x = its current LMArena Elo, y = its blended list price ($/M tokens, log). Coloured by quarter of the price observation. The cloud shifts down (same capability, cheaper) and right (new, smarter models arrive) over time. Hover any point for the model & price.
Price-performance frontier: the cheapest non-free blended price ($/M) to reach each Elo threshold, monthly. This is the right way to read "how cheap is a given capability level over time" — it declines cleanly as new, cheaper models cross each threshold (e.g. Elo ≥ 1400 fell from $26 → $0.11/M). Note: Elo here is each model's current score (no historical Elo available), so a tier only has data once models at that level exist. Free ($0) models excluded. (Earlier version showed a per-tier median, which rose because expensive new models entered the band — the frontier minimum is the honest metric.)