20% OFFon credit limits — Cohort-03 exclusiveClaim Offer →
k.
kapavamoda.buzz
private inference fabric
Sign In

Model Catalog — 49 endpoints • 4 surfaces

Production-ready • OpenAPI 3.1 • Fully managed • Search by endpoint ID

671B MoE max1M ctx residencyp50 TTFT 186ms

TEXT GENERATION

24 endpoints
GPT-5.6 Solcurrent
OpenAI • GPT-5.6 • gpt-5.6-sol
$2.50/$10.00
400K ctx
Modality
text+image → text
Architecture
Dense + reasoning
Performance
TTFT 320ms p50 • 145 tok/s
Best For
flagship reasoning, coding, agentic
— total— active
Feb 2026 flagship
POST /v1/inference/executex-kapa-key required
GPT-5.6 Terracurrent
OpenAI • GPT-5.6 • gpt-5.6-terra
$1.25/$6.00
400K ctx
Modality
text+image → text
Architecture
Dense
Performance
TTFT 210ms • 210 tok/s
Best For
balance intelligence/cost
POST /v1/inference/executex-kapa-key required
GPT-5.6 Lunacurrent
OpenAI • GPT-5.6 • gpt-5.6-luna
$0.50/$2.50
400K ctx
Modality
text+image → text
Architecture
Dense
Performance
TTFT 140ms • 340 tok/s
Best For
cost, high-volume
POST /v1/inference/executex-kapa-key required
GPT-5.4legacy
OpenAI • GPT-5 • gpt-5.4
$2.00/$8.00
400K ctx
Modality
text+image → text
Architecture
—
Performance
TTFT 380ms • 98 tok/s
Best For
hard reasoning, thinking-pro
POST /v1/inference/executex-kapa-key required
GPT-5.4 Minicurrent
OpenAI • GPT-5 • gpt-5.4-mini
$0.40/$1.80
400K ctx
Modality
text+image → text
Architecture
—
Performance
TTFT 160ms • 280 tok/s
Best For
subagents, batch
POST /v1/inference/executex-kapa-key required
gpt-oss-120Bcurrent
OpenAI • Open-OSS • gpt-oss-120b
open weights
128K ctx
Modality
text+image → text
Architecture
MoE 120B
Performance
TTFT 290ms • 110 tok/s
Best For
self-host, fine-tune
Apache 2.0
POST /v1/inference/executex-kapa-key required
Gemini 2.5 Procurrent
Google • Gemini 2.5 • gemini-2.5-pro
$1.25/$10.00
1M (2M via extension) ctx
Modality
text+image+audio+video → text
Architecture
Dense + adaptive thinking
Performance
TTFT 410ms • 124 tok/s
Best For
deep reasoning, 1M KV-cache residency
sunsets Oct 2026 -> 3.5
POST /v1/inference/executex-kapa-key required
Gemini 2.5 Flashcurrent
Google • Gemini 2.5 • gemini-2.5-flash
$0.30/$2.50
1M ctx
Modality
text+image+audio+video → text
Architecture
Dense
Performance
TTFT 180ms • 208 tok/s
Best For
low latency, agentic
POST /v1/inference/executex-kapa-key required
Gemini 3.7 Flashcurrent
Google • Gemini 3.x • gemini-3.7-flash
$1.50/$7.50
1M ctx
Modality
text+image+audio+video → text
Architecture
Dense
Performance
TTFT 160ms • 323 tok/s
Best For
multi-step, code, token-efficient
POST /v1/inference/executex-kapa-key required
Gemini 3.5 Flashcurrent
Google • Gemini 3.x • gemini-3.5-flash
$1.50/$9.00
1M ctx
Modality
text+image+audio+video → text
Architecture
Dense
Performance
TTFT 190ms • 201 tok/s
Best For
near-Pro at Flash cost
POST /v1/inference/executex-kapa-key required
Gemini 3.5 Flash-Litecurrent
Google • Gemini 3.x • gemini-3.5-flash-lite
$0.30/$2.50
1M ctx
Modality
text+image → text
Architecture
Dense
Performance
TTFT 95ms • 346 tok/s
Best For
subagent, high-volume
POST /v1/inference/executex-kapa-key required
Claude Fable 5current
Anthropic • Claude 5 • claude-fable-5
$10.00/$50.00
1M ctx
Modality
text+image → text
Architecture
Dense adaptive thinking always-on
Performance
TTFT 620ms • 42 tok/s
Best For
long-running agents, highest stakes
— active
slowest, most capable
POST /v1/inference/executex-kapa-key required
Claude Opus 5current
Anthropic • Claude 5 • claude-opus-5
$5.00/$25.00
1M ctx
Modality
text+image → text
Architecture
Dense adaptive
Performance
TTFT 480ms • 68 tok/s
Best For
complex agentic coding, enterprise
POST /v1/inference/executex-kapa-key required
Claude Sonnet 5current
Anthropic • Claude 5 • claude-sonnet-5
$2.00/$10.00
1M ctx
Modality
text+image → text
Architecture
Dense adaptive
Performance
TTFT 240ms • 145 tok/s
Best For
default production, balance
POST /v1/inference/executex-kapa-key required
Claude Haiku 4.5current
Anthropic • Claude 4.5 • claude-haiku-4-5-20251001
$1.00/$5.00
200K ctx
Modality
text+image → text
Architecture
Dense extended thinking
Performance
TTFT 120ms • 280 tok/s
Best For
fastest, classification, routing
POST /v1/inference/executex-kapa-key required
Claude Opus 4.8legacy
Anthropic • Claude 4 • claude-opus-4-8
$5.00/$25.00
1M ctx
Modality
text+image → text
Architecture
—
Performance
TTFT 520ms • 55 tok/s
Best For
legacy pin, 300K batch
POST /v1/inference/executex-kapa-key required
DeepSeek-V3current
DeepSeek • DeepSeek • deepseek-v3
$0.55/$2.19
128K ctx
Modality
text → text
Architecture
MoE MLA + DeepSeekMoE, FP8, MTP
Performance
TTFT 310ms • 185 tok/s
Best For
reasoning value, open weights
671B total37B activeMIT
POST /v1/inference/executex-kapa-key required
DeepSeek-R1current
DeepSeek • DeepSeek • deepseek-r1
$0.55/$2.19
128K ctx
Modality
text → text
Architecture
MoE reasoning via GRPO
Performance
TTFT 540ms • 72 tok/s
Best For
reasoning baseline, distilled 7B
671B total37B active
POST /v1/inference/executex-kapa-key required
Qwen3-235B-A22Bcurrent
Alibaba • Qwen3 • qwen3-235b-a22b
$1.00/$4.00
128K ctx
Modality
text → text
Architecture
MoE, thinking/non-thinking toggle
Performance
TTFT 360ms • 110 tok/s
Best For
multilingual 100+, C-Eval top
235B total22B activeApache 2.0
POST /v1/inference/executex-kapa-key required
Qwen3-Coder-480B-A35Bcurrent
Alibaba • Qwen3 • qwen3-coder-480b-a35b
$1.20/$4.50
256K (1M extrap.) ctx
Modality
text → text
Architecture
MoE coder, repo-scale
Performance
TTFT 420ms • 88 tok/s
Best For
agentic coding, front-end
480B total35B active
POST /v1/inference/executex-kapa-key required
Kimi K2current
Moonshot • Kimi • kimi-k2
$0.60/$2.50
128K (2M via 2.5) ctx
Modality
text → text
Architecture
MoE MLA 384 experts/8 active, MuonClip
Performance
TTFT 380ms • 95 tok/s
Best For
lossless long-context, agentic
1.04T total32B active
POST /v1/inference/executex-kapa-key required
GLM-4.5current
Zhipu • GLM • glm-4.5
$0.50/$1.50
200K ctx
Modality
text → text
Architecture
MoE GQA 96 heads, deep layers
Performance
TTFT 340ms • 102 tok/s
Best For
tool calling BFCL 77.8%, enterprise JSON
355B total32B active
POST /v1/inference/executex-kapa-key required
GLM-4.5-Aircurrent
Zhipu • GLM • glm-4.5-air
$0.30/$1.00
200K ctx
Modality
text → text
Architecture
MoE
Performance
TTFT 210ms • 165 tok/s
Best For
small agent, open variant
106B total12B active
POST /v1/inference/executex-kapa-key required
MiniMax abab7current
MiniMax • MiniMax • minimax-abab7
$0.80/$3.00
256K ctx
Modality
text → text
Architecture
Dense
Performance
TTFT 280ms • 130 tok/s
Best For
voice, multimodal product
POST /v1/inference/executex-kapa-key required

IMAGE GENERATION

10 endpoints
Imagen 4 Ultracurrent
Google • Imagen • imagen-4.0-ultra-generate
$0.04-0.06 /img
— ctx
Modality
text → image
Architecture
Diffusion
Performance
6s
Best For
photorealism, skin/hair, camera-language
POST /v1/inference/executex-kapa-key required
GPT Image 1.5current
OpenAI • GPT Image • gpt-image-1.5
$0.07 /HD img
— ctx
Modality
text+image → image
Architecture
Autoregressive
Performance
12s
Best For
prompt adherence, multi-subject, 99% text
POST /v1/inference/executex-kapa-key required
Flux 2 Procurrent
Black Forest Labs • Flux • flux-2-pro
$0.025-0.03 /img
— ctx
Modality
text+image → image
Architecture
Rectified Flow Transformer 32B
Performance
4s
Best For
cost/speed, aesthetic, open-weights friendly
POST /v1/inference/executex-kapa-key required
Flux 1.1 Pro Ultracurrent
Black Forest Labs • Flux • flux-1.1-pro-ultra
$0.04-0.06 /img
— ctx
Modality
text → image
Architecture
Rectified flow 12B
Performance
4.5s
Best For
photoreal value, self-host schnell Apache2.0
POST /v1/inference/executex-kapa-key required
Stable Diffusion 3.5 Largecurrent
Stability AI • Stable Diffusion • stable-diffusion-3.5-large
free self-host
— ctx
Modality
text+image → image
Architecture
Diffusion Transformer 8B
Performance
8-15s RTX4090
Best For
fine-tune, LoRA, ControlNet, local
POST /v1/inference/executex-kapa-key required
Stable Diffusion 4current
Stability AI • Stable Diffusion • stable-diffusion-4
free self-host OpenRAIL-M
— ctx
Modality
text+image → image
Architecture
Diffusion Transformer 14B
Performance
8-15s 16GB VRAM
Best For
14B local, 37% hyperparam search cut
POST /v1/inference/executex-kapa-key required
Recraft V4.1current
Recraft • Recraft • recraft-v4.1
$0.04 /img
— ctx
Modality
text → image+SVG
Architecture
Diffusion vector
Performance
5s
Best For
design systems, editable SVG
POST /v1/inference/executex-kapa-key required
Ideogram 4current
Ideogram • Ideogram • ideogram-4
$0.03 /img
— ctx
Modality
text → image
Architecture
Diffusion
Performance
5s
Best For
typography, posters, open weights
POST /v1/inference/executex-kapa-key required
Midjourney V8.1current
Midjourney • Midjourney • midjourney-v8.1
$10-120/mo sub
— ctx
Modality
text+image → image
Architecture
Proprietary
Performance
30-60s
Best For
art direction, Style DNA
POST /v1/inference/executex-kapa-key required
Seedream 5.0current
ByteDance • Seedream • seedream-5-lite
via ModelArk
— ctx
Modality
text → image
Architecture
Diffusion
Performance
5s
Best For
cinematic, semantic prompt
POST /v1/inference/executex-kapa-key required

VIDEO GENERATION

9 endpoints
Veo 3.1current
Google • Veo • veo-3.1
$0.03-0.15/s Lite/Full
— ctx
Modality
text+image → video+audio
Architecture
Diffusion + synced audio
Performance
30-60s/60s clip
Best For
synced dialogue, cinematic, Vertex SLA
POST /v1/inference/executex-kapa-key required
Runway Gen-4.5current
Runway • Runway • runway-gen-4.5
$12-76/mo
— ctx
Modality
text+image+video → video
Architecture
Diffusion + control surface
Performance
45s/10s
Best For
motion brush, inpainting, benchmark #1
POST /v1/inference/executex-kapa-key required
Kling 3.0current
Kuaishou • Kling • kling-3.0
$0.10/s $7.99/mo
— ctx
Modality
text+image → video+audio
Architecture
Diffusion + Omni lip-sync
Performance
40s/15s
Best For
value, 2min cont., image-to-video realism
POST /v1/inference/executex-kapa-key required
Luma Ray 3.2current
Luma • Dream Machine • luma-ray-3.2
$7.99-30/mo
— ctx
Modality
text+image+video → video
Architecture
Diffusion video-to-video
Performance
35s/10s
Best For
keyframed transform, HDR 16-bit EXR
POST /v1/inference/executex-kapa-key required
Wan 2.7current
Alibaba • Wan • wan-2.7
open weights self-host
— ctx
Modality
text+image → video
Architecture
Diffusion open weights
Performance
—
Best For
self-host, data-sovereign
POST /v1/inference/executex-kapa-key required
Hailuo 2.3current
MiniMax • Hailuo • hailuo-2.3
$9.99/mo
— ctx
Modality
text+image → video
Architecture
Diffusion
Performance
30s/10s
Best For
micro-expressions, budget
POST /v1/inference/executex-kapa-key required
Omni Flashcurrent
Google • Gemini Omni • omni-flash
$0.06/s
— ctx
Modality
text+image+video → video+audio
Architecture
Multimodal Omni Native
Performance
8s/8s clip
Best For
character-consistent video, up to 14 refs, SynthID
POST /v1/inference/executex-kapa-key required
Omni Flash 1.1current
Google • Gemini Omni • omni-1.1-flash
$0.08/s
— ctx
Modality
text+image+video → video+audio
Architecture
Omni 1.1 Multimodal
Performance
10s/8s clip
Best For
latest Omni, reasoning-enhanced video, 4K
POST /v1/inference/executex-kapa-key required
Sora 2 Prodeprecated
OpenAI • Sora • sora-2-pro
$0.75/s
— ctx
Modality
text+image → video+audio
Architecture
Diffusion Transformer
Performance
—
Best For
DEPRECATED - sunsets Sep 24 2026
POST /v1/inference/executex-kapa-key required

AUDIO GENERATION

6 endpoints
ElevenLabs TTS v2current
ElevenLabs • ElevenLabs • eleven-tts
$5-99/mo
— ctx
Modality
text → audio
Architecture
Neural TTS 32 langs
Performance
150ms
Best For
voice synthesis, dubbing, cloning
POST /v1/inference/executex-kapa-key required
Fish Audio S2 Procurrent
Fish Audio • Fish • fish-s2-pro
$15/1M chars
— ctx
Modality
text → audio
Architecture
4B 80+ langs
Performance
<150ms
Best For
cost-effective multilingual
POST /v1/inference/executex-kapa-key required
Suno v5.5current
Suno • Suno • suno-v5.5
$10-30/mo
— ctx
Modality
text → audio
Architecture
MusicGen 44.1kHz
Performance
8s/2min song
Best For
viral songs, vocals 9.5/10
POST /v1/inference/executex-kapa-key required
Udio v2current
Udio • Udio • udio-v2
$10-30/mo
— ctx
Modality
text+audio → audio
Architecture
Diffusion 48kHz
Performance
15s
Best For
studio-grade, inpainting, instrumental
POST /v1/inference/executex-kapa-key required
Stable Audio 3 Largecurrent
Stability AI • Stable Audio • stable-audio-3-large
open weights
— ctx
Modality
text+audio → audio
Architecture
Diffusion 2.7B
Performance
2s H200
Best For
local Mac M4, SFX
POST /v1/inference/executex-kapa-key required
ElevenLabs Musiccurrent
ElevenLabs • ElevenLabs • eleven-music
$5-330/mo
— ctx
Modality
text → audio
Architecture
Licensed catalog
Performance
10s
Best For
commercial-safe, licensed Kobalt/Merlin
POST /v1/inference/executex-kapa-key required
Note: All model endpoints are versioned and maintained by our team. Deprecated IDs remain with red badge for migration tracking — do not route production traffic.