GetToken · Model Plaza

Model Plaza

Every model across every channel, with transparent pricing and real request monitoring. Price = official rate × channel multiplier, no hidden fees; CNY prices are compared against official pricing so savings are clear at a glance.

OperationalUpdated today 23:26 · refreshes every 60sHealth from real requests (7d)Consumption baselinegpt-6-astra
All channels
5
Models
29
Avg success rate (7d)
58.6%
Avg TTFT P50
32500ms
O

国模特惠

— · 5 models
×0.60
ModelInput (ours)Output (ours)Cache readMultiplierConsumptionOfficial (in/out)vs officialHealth · 7d
deepseek-v4.1-flash
openai · 国模特惠
$0.090≈ ¥0.60 / 1M tok
$0.360≈ ¥2.41 / 1M tok
$0.0018≈ ¥0.01 / 1M tok
×0.60×0.03
¥3.00 / ¥9.00peak · off-peak ¥1.5 / ¥4.5
Save 77%— no monitor data
glm-5.3
openai · 国模特惠
$0.840≈ ¥5.63 / 1M tok
$2.64≈ ¥17.69 / 1M tok
$0.156≈ ¥1.05 / 1M tok
×0.60×0.19
$1.40 / $4.40≈ ¥1.40 / ¥4.40 · 1M tok
Save 40%— no monitor data
glm-5.3-flash
openai · 国模特惠
$0.090≈ ¥0.60 / 1M tok
$0.300≈ ¥2.01 / 1M tok
$0.018≈ ¥0.12 / 1M tok
×0.60×0.02
$0.15 / $0.50≈ ¥0.15 / ¥0.50 · 1M tok
Save 40%— no monitor data
glm-5.3-flashx
openai · 国模特惠
$0.090≈ ¥0.60 / 1M tok
$0.300≈ ¥2.01 / 1M tok
$0.018≈ ¥0.12 / 1M tok
×0.60×0.02
$0.15 / $0.50≈ ¥0.15 / ¥0.50 · 1M tok
Save 40%— no monitor data
kimi-k3
openai · 国模特惠
$1.80≈ ¥12.06 / 1M tok
$9.00≈ ¥60.30 / 1M tok
$0.180≈ ¥1.21 / 1M tok
×0.60×0.60
$3.00 / $15.00≈ ¥3.00 / ¥15.00 · 1M tok
Save 40%— no monitor data
A

Claude-Kiro-特惠

特惠渠道 · 3 models
×1.00
ModelInput (ours)Output (ours)Cache readMultiplierConsumptionOfficial (in/out)vs officialHealth · 7d
claude-fable-5
anthropic · Claude-Kiro-特惠
$10.00≈ ¥67.00 / 1M tok
$50.00≈ ¥335 / 1M tok
$1.00≈ ¥6.70 / 1M tok
×1.00×0.80
$10.00 / $50.00≈ ¥10.00 / ¥50.00 · 1M tok
Same as official— no monitor data
claude-fable-5-1
anthropic · Claude-Kiro-特惠
$10.00≈ ¥67.00 / 1M tok
$50.00≈ ¥335 / 1M tok
$0.250≈ ¥1.68 / 1M tok
×1.00×0.80
$10.00 / $50.00≈ ¥10.00 / ¥50.00 · 1M tok
Same as official— no monitor data
claude-opus-5-5
anthropic · Claude-Kiro-特惠
$4.00≈ ¥26.80 / 1M tok
$20.00≈ ¥134 / 1M tok
$0.200≈ ¥1.34 / 1M tok
×1.00×0.60
$4.00 / $20.00≈ ¥4.00 / ¥20.00 · 1M tok
Same as official— no monitor data
A

Claude稳定

稳定渠道 · 3 models
×1.00
ModelInput (ours)Output (ours)Cache readMultiplierConsumptionOfficial (in/out)vs officialHealth · 7d
claude-fable-5
anthropic · Claude稳定
$10.00≈ ¥67.00 / 1M tok
$50.00≈ ¥335 / 1M tok
$1.00≈ ¥6.70 / 1M tok
×1.00×2.40
$10.00 / $50.00≈ ¥10.00 / ¥50.00 · 1M tok
Same as official— no monitor data
claude-fable-5-1
anthropic · Claude稳定
$10.00≈ ¥67.00 / 1M tok
$50.00≈ ¥335 / 1M tok
$0.250≈ ¥1.68 / 1M tok
×1.00×2.40
$10.00 / $50.00≈ ¥10.00 / ¥50.00 · 1M tok
Same as official— no monitor data
claude-opus-5-5
anthropic · Claude稳定
$4.00≈ ¥26.80 / 1M tok
$20.00≈ ¥134 / 1M tok
$0.200≈ ¥1.34 / 1M tok
×1.00×0.96
$4.00 / $20.00≈ ¥4.00 / ¥20.00 · 1M tok
Same as official— no monitor data
O

OpenAI特惠

— · 11 models
×1.00
ModelInput (ours)Output (ours)Cache readMultiplierConsumptionOfficial (in/out)vs officialHealth · 7d
codex-auto-review
openai · OpenAI特惠
$0.200≈ ¥1.34 / 1M tok
$1.20≈ ¥8.04 / 1M tok
$0.020≈ ¥0.13 / 1M tok
×1.00×0.02
$0.20 / $1.20≈ ¥0.20 / ¥1.20 · 1M tok
Same as official— no monitor data
gpt-5.6
openai · OpenAI特惠
$5.00≈ ¥33.50 / 1M tok
$30.00≈ ¥201 / 1M tok
$0.500≈ ¥3.35 / 1M tok
×1.00×0.58
$5.00 / $30.00≈ ¥5.00 / ¥30.00 · 1M tok
Same as official— no monitor data
gpt-5.6-luna
openai · OpenAI特惠
$0.200≈ ¥1.34 / 1M tok
$1.20≈ ¥8.04 / 1M tok
$0.020≈ ¥0.13 / 1M tok
×1.00×0.02
$0.20 / $1.20≈ ¥0.20 / ¥1.20 · 1M tok
Same as official— no monitor data
gpt-5.6-sol
openai · OpenAI特惠
$5.00≈ ¥33.50 / 1M tok
$30.00≈ ¥201 / 1M tok
$0.500≈ ¥3.35 / 1M tok
×1.00×0.58
$5.00 / $30.00≈ ¥5.00 / ¥30.00 · 1M tok
Same as official— no monitor data
gpt-5.6-terra
openai · OpenAI特惠
$2.00≈ ¥13.40 / 1M tok
$12.00≈ ¥80.40 / 1M tok
$0.200≈ ¥1.34 / 1M tok
×1.00×0.23
$2.00 / $12.00≈ ¥2.00 / ¥12.00 · 1M tok
Same as official— no monitor data
gpt-6-astra
openai · OpenAI特惠
$10.00≈ ¥67.00 / 1M tok
$50.00≈ ¥335 / 1M tok
$1.00≈ ¥6.70 / 1M tok
×1.00×1.00
$10.00 / $50.00≈ ¥10.00 / ¥50.00 · 1M tok
Same as official— no monitor data
gpt-6-sol
openai · OpenAI特惠
$2.00≈ ¥13.40 / 1M tok
$10.00≈ ¥67.00 / 1M tok
$0.200≈ ¥1.34 / 1M tok
×1.00×0.20
$2.00 / $10.00≈ ¥2.00 / ¥10.00 · 1M tok
Same as official— no monitor data
gpt-6.1-sol
openai · OpenAI特惠
$2.00≈ ¥13.40 / 1M tok
$10.00≈ ¥67.00 / 1M tok
$0.100≈ ¥0.67 / 1M tok
×1.00×0.20
$2.00 / $10.00≈ ¥2.00 / ¥10.00 · 1M tok
Same as official— no monitor data
gpt-image-2
openai · OpenAI特惠
$5.00≈ ¥33.50 / 1M tok
$10.00≈ ¥67.00 / 1M tok
$1.25≈ ¥8.38 / 1M tok
×1.00×0.50
$5.00 / $10.00≈ ¥5.00 / ¥10.00 · 1M tok
Same as official— no monitor data
gpt-image-2.5-flare
openai · OpenAI特惠
$5.00≈ ¥33.50 / 1M tok
—
$1.25≈ ¥8.38 / 1M tok
×1.00×0.50
$5.00 / —≈ ¥5.00 / — · 1M tok
Same as official— no monitor data
gpt-image-2.5-sunburst
openai · OpenAI特惠
$5.00≈ ¥33.50 / 1M tok
—
$1.25≈ ¥8.38 / 1M tok
×1.00×0.50
$5.00 / —≈ ¥5.00 / — · 1M tok
Same as official— no monitor data
O

OpenAI稳定

— · 7 models
×1.00
ModelInput (ours)Output (ours)Cache readMultiplierConsumptionOfficial (in/out)vs officialHealth · 7d
codex-auto-review
openai · OpenAI稳定
$0.200≈ ¥1.34 / 1M tok
$1.20≈ ¥8.04 / 1M tok
$0.020≈ ¥0.13 / 1M tok
×1.00×0.03
$0.20 / $1.20≈ ¥0.20 / ¥1.20 · 1M tok
Same as official
No data
Success 100.0%P50 60000ms11 tok/s
gpt-5.6
openai · OpenAI稳定
$5.00≈ ¥33.50 / 1M tok
$30.00≈ ¥201 / 1M tok
$0.500≈ ¥3.35 / 1M tok
×1.00×0.76
$5.00 / $30.00≈ ¥5.00 / ¥30.00 · 1M tok
Same as official
Critical
Success 0.0%P50 —
gpt-5.6-luna
openai · OpenAI稳定
$0.200≈ ¥1.34 / 1M tok
$1.20≈ ¥8.04 / 1M tok
$0.020≈ ¥0.13 / 1M tok
×1.00×0.03
$0.20 / $1.20≈ ¥0.20 / ¥1.20 · 1M tok
Same as official
Critical
Success 0.7%P50 15000ms14 tok/s
gpt-5.6-sol
openai · OpenAI稳定
$5.00≈ ¥33.50 / 1M tok
$30.00≈ ¥201 / 1M tok
$0.500≈ ¥3.35 / 1M tok
×1.00×0.76
$5.00 / $30.00≈ ¥5.00 / ¥30.00 · 1M tok
Same as official
Critical
Success 83.3%P50 30000ms20 tok/s
gpt-5.6-terra
openai · OpenAI稳定
$2.00≈ ¥13.40 / 1M tok
$12.00≈ ¥80.40 / 1M tok
$0.200≈ ¥1.34 / 1M tok
×1.00×0.30
$2.00 / $12.00≈ ¥2.00 / ¥12.00 · 1M tok
Same as official
Critical
Success 70.7%P50 30000ms17 tok/s
gpt-6-astra
openai · OpenAI稳定
$10.00≈ ¥67.00 / 1M tok
$50.00≈ ¥335 / 1M tok
$1.00≈ ¥6.70 / 1M tok
×1.00×1.30
$10.00 / $50.00≈ ¥10.00 / ¥50.00 · 1M tok
Same as official
Critical
Success 78.9%P50 30000ms17 tok/s
gpt-6.1-sol
openai · OpenAI稳定
$2.00≈ ¥13.40 / 1M tok
$10.00≈ ¥67.00 / 1M tok
$0.100≈ ¥0.67 / 1M tok
×1.00×0.26
$2.00 / $10.00≈ ¥2.00 / ¥10.00 · 1M tok
Same as official
Warning
Success 76.5%P50 30000ms20 tok/s

✕ Multipliers · How the consumption multiplier works

Channel multiplier
Each channel (group) has a base multiplier applied to its models. Lower is cheaper; 1.00 means 1:1 with the reference rate.
Image / video multipliers
Some channels apply separate multipliers to image or video models (e.g. 1.0 / 1.5), independent of text.
Official price comparison
Our unit prices are always shown in USD (the billing basis) with a CNY reference at a fixed rate of 1 USD = ¥6.7. Official prices are shown in their native currency: USD for international models, CNY for domestic CNY-priced ones (e.g. DeepSeek, which uses peak/off-peak pricing — peak 9:00–12:00 / 14:00–18:00 CST, off-peak at half price; comparisons use peak prices).
How the discount is computed
Discount = our blended effective unit price ÷ official unit price (input:output = 3:1). Both sides share the same currency basis, so the ratio is exchange-rate independent. Billing remains in USD credits; CNY figures are for comparison only.
Consumption multiplier = our charge ÷ baseline-model charge for identical input / output tokens·The baseline model is configured in the admin console (current: gpt-6-astra). If the baseline costs $1.00 for the same tokens and we charge $0.80, the consumption multiplier is ×0.80. Channel multiplier and consumption multiplier are independent: the former is a billing parameter, the latter is relative to the baseline model.

FAQ

Is billing actually in CNY?

No. Actual billing is in USD credits, so our unit prices are always shown in USD (CNY at a fixed 1 USD = ¥6.7 is reference only). Official prices are shown in their native currency: USD for international models, CNY for domestic CNY-priced ones. Always refer to your account billing records.

What's the difference between channel multiplier and consumption multiplier?

The channel multiplier is a billing parameter: model price × multiplier = what you pay here. The consumption multiplier is relative to the baseline model: with identical input/output tokens, your cost here as a multiple of the baseline model's cost (e.g. ×0.40 means you pay 40% of what the baseline would cost). They are independent of each other.

How is the discount vs official pricing calculated?

Using a blended weight of 75% input + 25% output, our effective price divided by the converted official price. Both sides are the same currency and basis, so the exchange rate cancels out — the discount is rate-independent. Official prices follow published rates.

Why does DeepSeek show two prices?

DeepSeek's official pricing is time-based: full price during peak hours (9:00–12:00, 14:00–18:00 CST) and roughly half price otherwise. The official column shows both tiers; the discount is computed against the peak price as a conservative estimate — off-peak is even cheaper.

Where does the monitoring data come from?

Entirely from real requests over the past 7 days, aggregated hourly: TTFT P50 latency, success rate, and generation throughput. No simulated data — models without recent traffic show "no data".

Why does the consumption multiplier show "baseline not set"?

The consumption multiplier needs a baseline model as the 1× anchor (equal cost for equal tokens). Until a baseline is configured in the admin console, the column shows a placeholder; it lights up automatically once configured.

Data refreshes every 60s · monitoring from real requests over the last 7 days (TTFT P50 / success rate / throughput)
Our unit prices are always in USD (billing basis); CNY uses a fixed rate of 1 USD = ¥6.7 for reference only · Official prices are shown in their native currency (USD for international, CNY for domestic CNY-priced models); DeepSeek uses peak/off-peak prices (peak 9:00–12:00 / 14:00–18:00 CST)