Token Demand Index

Name: Token Demand Index (AI inference consumption gauge)
Creator: Bargo
License: https://www.bargo.ai/terms

Tracked inference demand reached 116.4 trillion tokens per week as of 2026-09-27, driving the Token Demand Index to 494.9, +67% over 30 days and +395% since 2026-05-06, while the demand-weighted effective price sits at $0.72 per 1M tokens. Open-source models account for 74.7% of tracked tokens, confirming that falling unit prices reflect surging consumption rather than margin collapse as the market grows larger. It is the output-side companion to the Compute Tightness Index (the GPU input).

497Demand index (100 = 2026-05-06)

116.9TTokens / week

▲ +66%Demand, 30 days

73.3%Open-source share

$0.73Effective $/1M

$85.0MImplied spend/week*

As of 2026-09-28, tracked AI inference runs at 116.9T tokens/week — up +66% in 30 days and +397% since 2026-05-06. Open-source carries 73.3% of those tokens, holding the demand-weighted price near $0.73/1M — a fraction of frontier list prices even as frontier models themselves got pricier.

Open vs closed-model pricing

Blended price per million tokens. The closed series demand-weights each named provider group's median price by its trailing-seven-day token volume.

$0.47Open-weight models · $/1M tokens · 73.3% of tracked demand

$2.92Closed models · $/1M tokens · 7.2% of tracked demand

Open-weight $/1MClosed-model $/1M

Open-weight = the Open-source provider group. Closed = Claude, OpenAI, Google, and xAI. The unclassified “Other” bucket is excluded from the closed-model price.

⤓ Download full daily series (CSV)

Token demand vs the effective price of inference

Total tokens/weekEffective $/1M (right)

Who consumes the tokens

Open-sourceOtherClaudeOpenAIGooglexAI

By provider group — latest snapshot

Click a group to expand its top contributing models.

Group	Blended $/1M	Tokens/week	Share	30d demand
▸Open-source	$0.47	85.7T	73.3%	▲ +67%
deepseek/deepseek-v4.1-flash		20.8T	24.3% of group
z-ai/glm-5.3-flash		13.3T	15.5% of group
tencent/hy4-preview		9.0T	10.5% of group
deepseek/deepseek-v4-flash		7.6T	8.8% of group
xiaomi/mimo-v2.6-flash		6.9T	8.1% of group
deepseek/deepseek-v4-flash		3.3T	3.8% of group
z-ai/glm-5.3		2.7T	3.2% of group
tencent/hy3		2.5T	2.9% of group
xiaomi/mimo-v2.5		2.0T	2.3% of group
meta/muse-spark-1.3-contributor		1.6T	1.8% of group
▸Other	$0.88	22.8T	19.5%	▲ +906%
stealth/space-bunny-alpha		18.2T	79.7% of group
typesafe/jev-1.13		2.7T	11.8% of group
upstage/solar-pro4		1.5T	6.8% of group
upstage/solar-mini4		105.0B	0.5% of group
voyageai/rerank-2.5-lite		54.1B	0.2% of group
inception/mercury-2.5		46.8B	0.2% of group
poolside/laguna-s-2.1		41.2B	0.2% of group
thinkingmachines/inkling		26.4B	0.1% of group
voyageai/rerank-2.5		25.7B	0.1% of group
amazon/nova-micro-v1		20.9B	0.1% of group
▸Claude	$6.00	2.0T	1.7%	▼ -30%
anthropic/claude-opus-5		820.7B	41.1% of group
anthropic/claude-4.6-sonnet		336.0B	16.8% of group
anthropic/claude-4.8-opus		257.4B	12.9% of group
anthropic/claude-4.5-haiku		243.0B	12.2% of group
anthropic/claude-4.6-opus		116.5B	5.8% of group
anthropic/claude-4.5-sonnet		101.2B	5.1% of group
anthropic/claude-4.7-opus		50.9B	2.6% of group
anthropic/claude-5-fable		43.7B	2.2% of group
anthropic/claude-4-sonnet		13.7B	0.7% of group
anthropic/claude-4.5-opus		11.5B	0.6% of group
▸OpenAI	$2.41	3.9T	3.3%	▼ -65%
openai/gpt-5.6-sol		1.7T	44.3% of group
openai/gpt-5.6-terra		502.8B	13% of group
openai/gpt-4o-mini		234.4B	6.1% of group
openai/gpt-5.6-luna-pro		214.9B	5.6% of group
openai/text-embedding-3-small		214.6B	5.6% of group
openai/gpt-5-mini-2025-08-07		124.2B	3.2% of group
openai/gpt-5.4-nano		105.9B	2.7% of group
openai/gpt-4.1-mini-2025-04-14		91.6B	2.4% of group
openai/gpt-5.4-mini		90.1B	2.3% of group
openai/gpt-4.1-nano-2025-04-14		67.4B	1.7% of group
▸Google	$1.13	1.7T	1.5%	▼ -17%
google/gemini-2.5-flash-lite		705.8B	40.5% of group
google/gemini-3.7-flash		483.7B	27.7% of group
google/gemini-3.1-pro-preview		203.3B	11.7% of group
google/gemini-3.6-flash		146.9B	8.4% of group
google/gemini-3.1-flash-lite-preview		91.6B	5.2% of group
google/gemini-2.5-pro		63.6B	3.6% of group
google/gemini-embedding-2		15.2B	0.9% of group
google/gemini-embedding-001		9.1B	0.5% of group
google/gemini-3.1-flash-lite		5.4B	0.3% of group
google/gemini-embedding-2-preview		3.4B	0.2% of group
▸xAI	$1.56	774.6B	0.7%	▲ +11%
x-ai/grok-4.7		394.5B	50.9% of group
x-ai/grok-4.6		215.5B	27.8% of group
x-ai/grok-4.5		69.7B	9% of group
x-ai/grok-4.3		59.9B	7.7% of group
x-ai/grok-4.20		30.8B	4% of group
x-ai/grok-build-0.1		2.2B	0.3% of group
x-ai/grok-4.20-multi-agent		1.1B	0.1% of group
x-ai/grok-imagine-image-2.0		544.1M	0.1% of group
x-ai/grok-imagine-image-quality		406.1M	0.1% of group

Open-source = aggregated weights-available labs (DeepSeek, Qwen, Llama, Mistral, etc.). "Other" = unclassified OpenRouter providers. Share = % of tracked tokens. *Implied spend = list price × tokens, not realized revenue.

Methodology

Built from OpenRouter's model rankings + pricing, grouped into Claude / OpenAI / Google / xAI / Open-source / Other. Token counts are OpenRouter's trailing-7-day figures (the rankings "week" view) — i.e. each point is tokens processed over the prior week, not a single day. Per group: blended median $/1M (75/25 input/output) and weekly total_tokens. Demand index = latest weekly total ÷ the first week's total × 100. Effective $/1M = demand-weighted blended price (as-of prices carried forward across snapshots where only one side updated). Series since 2026-05-06; it grows as new snapshots land. Full history as CSV. Coverage is OpenRouter traffic — a large but partial slice of the market.

Pair it with the Compute Tightness Index (GPU input) for both sides of the AI-compute economy. Query the raw data from an agent via the Bargo MCP endpoint (get_inference_economics).